How would an Ent evolve. Very slowly, I assume - maybe something can be done with some interesting plants, a handful of coincidences, and a lot of time.
In any case, since of the traits we want, mobile is probably the rarest (among those that actually exist), I thought of tumbleweeds, and of carnivorous plants, and even the sensitive plant. That might be a good place to start. Of those, I picked the carnivorous plants, like the Venus flytrap or Waterwheel plant, or maybe even a Bladderwort. These are active traps, where the plant is triggered and moves to catch the prey, so it has some sensory feedback and movement already in its repertoire (unlike passive traps like flypaper, pitfall, or lobster-pot traps). Carnivorous plants are also known to grow in poor soils, one of the reasons that they turn to carnivory in the first place.
So, we have a carnivorous plant, and it is already sensitive, and a little bit mobile. Since catching its prey is a major investment of energy - it needs to grow shut, and too many missed snaps, or even too many feedings, can overreach its resources and it dies - it is already on the path to increase its sensory capabilities as much as it can. Since it already habituated to poor soils and difficult growing habitats, it is a good candidate for a flying plant - that is, growing with its roots not in the ground (on an outcropping of rock, or on dead wood, or on the living branches of another plant - not flying, flying, ok?).
Somehow, in some way, the plant (I picked the Venus flytrap, to be specific) ends up in more fertile ground - dropped by some bird, or carried by an animal. Just to up our chances, we should design a geographic proximity so that this happens often enough for a while, perhaps a raised bog with rock outcroppings separating it a more fertile growing ground below, with maybe some bird looking there for twigs for its nest, or something - since our plant will like both the bog and the outcroppings, and it might find the valley right for evolving in. When the seeds or cuttings land, the places most like its natural habitat, and therefore where it is likely to germinate, are the places already mentioned as "flying", the outcroppings of rocks and branches.
Of those, those which end up in the branches of a living tree have the opportunity to change in a new way, to adapt to their new growing condition. With sudden proximity to nutrients in plant-accessible form, in the tree next to them, some may find a way to grow partially parasitic (hemiparasitic, to be specific). Parasitic plants have evolved before, so it isn't impossible - especially since our plant will need to get water and mineral nutrients from somewhere, if the amount that germinated it was a thin layer on a living branch. If it can figure out any mechanism to tap into the branch, it may find itself a successful symbiote - since it may trap insects harmful to its host, and will not have to face much competition from its own kind in tree branches. This was one reason to pick the Venus flytrap as our base, since it specializes in slightly larger prey like beetles, which might let it avoid necessary pollinators (unlike species targeting smaller aerial insects as prey).
So, we have a carnivorous, semi-parasitic plant. With the slow adaptation of its roots to feed from the host plant, our plant now has resources - resources it lacked in its environment of poor growing soil - to evolve some more. Perhaps it can increase its sensory capabilities, and centralize its ability to act on those senses - the better to select the right kind of prey, and to prevent wasting energy on falsely triggered snaps. Perhaps it should tweak its reproductive strategies to better colonize its new growing grounds, to keep the species in the trees. Since it is already partially mobile, maybe it can take advantage of that, instead of trying to make fruit attractive to birds or whatnot to propagate its seeds high off the ground.
So, a tumbleweed strategy. The tumbleweeds are adapted to windy plains, and spread widely by shedding a ball (or bunch, whatever), of dead plant material, and let it be battered about by the wind, which slowly shakes out the seeds hidden inside and gives them a lot of spread. Obviously, our plant is in a forest, and has neither strong winds nor a flat plain - but the idea might be adapted. Maybe its seeds can be caught in a small, light bunch of dead twigs or shoots or even roots, which might be tossed to another tree by a breeze, or better yet grabbed by a bird for a nest. It will take a long time, of course, but the theory is sound. And the plant has some mobility and sensory capabilities, it might even be doing this on sort-of purpose eventually (sticking seeds on birds, or shaping twigs to carry them) after a while and quite a lot of selection towards activity (and sentience), of course.
Okay, now we have a parasitic plant mobile two ways! Actually, I think we're going to end up with tree sprites long before we get to ents, so bear with me. Mobility is expensive, so let's start putting that to work, not just the carnivorism. So, the plant, having adapted to tap into the host tree for water and minerals, will obviously want to stay on living material - if something is going on, if a branch is injured or sick, or even just the eating is better over that-a-way a little, the ability to move will be priceless. It can even move away from being eaten, or environmental hazards, or other sources of difficulty or injury, if it can move along the branch even a bit. Many parasitic plants propagate by runners and cuttings, with a long branch having the ability to put down new roots. If the old branch dries up and dies (to avoid overgrowing its host or its own abilities, or if the old spot stopped being viable), the plant can slowly, slowly, over months and years inch along branches to better feeding grounds.
Now, since our plant is a hemiparasite, it needs its host to be living. Trees with more active versions of our plant will therefore survive longer, since it will have fewer predatory insects damaging it with more coverage, and less of a parasitic burden on it with smaller plants - so more mobile versions will serve their trees best, and thus survive longer to continue to propagate. The tree may reward this behavior by developing nodes along the branches, around insect-areas, where the bark is thin to "encourage" the plant to root down here where more insects are, and after a time "dry up" the port where no insect damage is so that the plant keeps moving instead of overgrowing and demanding more resources. What tree is this? I don't even know, but it won't last forever, so go with it.
Mobile little plants, "racing" around their trees and munching on insects. We are actually getting there. With mobility and carnivority both encouraged, our plant has reason to continue along the path to better sensory capabilities, better mobility, and better reactionary capabilities... the same sorts of pressures that led to animals developing sentience, and eventually sapience in our humans. Whoo. In any case, now that we have mobile insect-munching plants, they will be sorta-competing with animal insectivores. Now, they won't ever be as quick as their animal counterparts, but what they can lean on is their plant heritage - they can tap directly into the tree, via rootlings, and take nourishment that way, and they can still photosynthesize for a bit more nourishment (encourage more leaf growth, perhaps, so they look more tree-ish - which is also camouflage). So to their advantage is multiple options, for when there aren't enough insects.
But the other thing they can develop, that will really help a lot? the ability to plan ahead. Especially since they will need to plan months ahead, to get things done (hey, this is sounding entish!). So, to go with their evolving sensory capabilities, they develop something like a neural network (can't be too specialized yet, since they are still moving by a grow-and-abandon strategy). This will let them react to stimuli more centrally, instead of just locally, let them correlate all sensory input and make moves based on that, and also lay the groundwork for future actual thinking capabilities.
So, being more mobile, and starting to be more capable of central, active decisions, our plant is adapting. One thing it will want, is to speed up and specialize its root system - tapping into branches where it pleases, where it wants to go, not just where it happens to land. At this point, if it had a tree species that adapted the node-system to nourish it and encourage mobility, it will no longer be necessary (though not abandoned, either - just that offspring can spread back to other species). Its leaves and even twigs might adapt to look more like it's tree, too, for camouflage, giving us "species" of ent later on.
We are into the home stretch, now. Recall that everything had to be decentralized, because our plant was going with a grow-and-abandon strategy of moving... well, the next step is to centralize. Instead of growing in a new direction and abandoning the growth placed elsewhere, the plant learns to "keep" and "re-purpose" a great deal of that effort - by losing less and less of the old growth until it is just abandoning the roots, and using its shiny mobility to "swing" the stems in the direction it wants to go, and put new roots down fast. Eventually, it may or may not reach the stage where it chooses short, spiked roots that can bite deep "quickly", and pull up without abandoning much of anything. Anyway, with this change comes the possibility of centralization. It can grow a node, where it can specialize sensory organs (animals evolved them, the attempts and failures would probably be recognizably parallel). It can centralize its reaction abilities, and decision making capabilities, into something vaguely analogous to a brain. It is still slow, it is still so very slow by animal standards - but now it can learn to speed up a little. It now has a lot more resources, since it isn't "abandoning" its leftover growth and starting new.
At this point, the plants will split into two populations (roughly. ish.) One sort will limit growth and continue to abandon overgrowth to stay small - recall, one of the reasons to be mobile was not to be too great a drain on the host tree by growing too large. Over time, they will develop first an instinct, then a culture, of protecting the mother-tree - since they are still far less mobile than animal insectivores, and their edge in competition against them encourages protecting and tending the host tree. Limiting their size will encourage the extra resources previously used for growth to be used elsewhere, letting them develop increasing mobility, speed, active defensive instincts, and eventually, some measure of intelligence (to animal levels, say). as small as they are, a tree may support many without strain - and being tribal (originally pack-oriented) will let them chase off larger dangers more efficiently, especially since they are still quite slow. Here are your tree sprites - small, lively, vicious in defense of their home-tree.
The other path, does not involve staying small. After all, it can move and react much faster than before, and adapt much quicker. In this variation, when the tree starts to show signs of weakness, from the burden of the growing plants - they jump trees. They will learn to root down quickly, to bite deep to take sap from a new tree, and to loosen their roots and move on when the tree falters. Moving from one tree to another lets them move on and stay alive even when they reach sizes that would be a burden to a single tree over time. They might stay with the same species (especially if they have developed a measure of camouflage), but they will quickly learn to move farther and farther, take bigger and bigger leaps, from branch to branch and tree to tree, even perhaps spanning and feeding from multiple trees at once when large enough. Once they're large enough, they will more precisely develop sturdier structural growth than they needed while small (from the woody stems of large vines, to something more like tree trunks themselves). Once they're large enough, and mobile enough, they might come down from the trees and move between them, at least long enough to get to a new tree to feed from.
Their instinct and culture will end up very different form the tree-sprites. Overlooking a number of trees will have them develop a more distant relationship - they won't develop the tree sprites' aggressive reliance on and devotion to their mother-tree, but they will care for the trees they feed from - tree herder, in truth - since that is their survival edge over their animal competitors (well, that and not being very edible except around the very edges). They will adapt to take on (or chase away) larger animals in defense of their trees, those that damage them, and even perhaps tear apart and eat them - they still are carnivorous plants. For that they would need to develop a larger mouth (mouth-flower, or something) for insects, the regular flowers will suffice. In any case, the more mobile they are, the fewer flowers they would have needed - since they are expensive to grow, and can move the flower to the prey instead of happening on it. This would have been part of the growth from flower-plant to treelike, I forgot to mention it earlier.
How big will they get? Not sure, but they will likely be quite spindly (to mimic branches), and can grow long and light because of it - and will probably keep growing most of their lives, until they are too big to support their needs or something else (injury, illness) kills them. Also, being plant based, might not need to support quite so much fleshy, squishy organs - long thin decentralized organs surrounded by structural growth will be better for their purposes. If their "feeding roots" come from the hands, they will be working with, not against, gravity for that part of their circulation system/feeding cycle. And they aren't likely to be "bipedal", so to speak, but to have several support structures (with no particular restriction on number) and be able to raise up to a height, probably on two "limbs" (kinda like bears) to reach up into their trees, so height when "standing" can be much greater than when otherwise moving. And given "several" limbs, people would likely try to equate them to something humanoid, to arms and legs, even if they might have five or six limbs, instead (though hub of sensory organs and nerve tissue will be "head" even if not very similar to our faces, people will still try). They may or may not mind minor parasites or flying plants (mosses or fungi), or might even cultivate them for some benefit - so there might be your "beards" or even "clothing".
But between the larger territory and the weaker connection to any individual tree, they will not develop as much possessive aggressiveness - instead, they will lean more on their hard-evolved planning ahead, allowing or even planning actions destructive to a single tree for the better growth of the whole area (or patch, or herd, whatever). Between that and the larger size, they are in a better position to eventually, eventually, develop something approaching sapience. They will be more likely to run into each other, being far more mobile, and need some language - and given they will not range far from their trees, a "taught" language will be more useful than an instinctive one, especially once they run into other "species" of ent (with different leaf-patterns and coloration, and thus a separation from the common ancestor at least from before tree sprites). It will be slow language, though, they are still living on plant time-scales. They can develop culture. And that, my friends, is how you grow an Ent.