Let's start with a size. Just for grins, assume the interior is a cylinder 1 km in diameter with a 10 km length (that's 6 miles long). It will have to rotate at a bit over 1 rpm to provide 1 g at the inner surface. Total surface area is about 30 km², or about 12 square miles.
Construction. Let's assume a final wall thickness of 100 meters. The total wall volume is about 3.6 km³, so the original asteroid had to be about 2 km in diameter. Absolute minimum power level required to melt this chunk of iron will be ~ 8 TW, since at melting point it will be radiating about 300 kW per square meter in blackbody radiation. At earth orbit, that will require a bit more than 5 million square kilometers of mirror. That's a circular mirror about 2500 km in diameter.
Lighting: To provide normal daylight levels (1 kW / m²) over the entire surface, the total power will be about 30 GW. Pushing this through a port in the end cap will be hazardous. If you assume a 100 m diameter window, the power levels at the window will be just about 1 MW / m², or 1000 times the brightness the sun. This beam will have to be sent down the axis with a series of mirrors all the way down the axis to spread the light out to the surface. So the axis will be very definitely off-limits. And frankly, I don't know quite how to specify a high-quality window 100 meters in diameter which will contain 10 to 15 psi. It will have to be one piece, I suspect, since if you make it out of panels the supporting structure will have to take high temperatures (due to the power flux) and still be strong enough to hold together under the pressure. Titanium/sapphire, perhaps? I'm not sure about cost for this project. Since the colony fabrication required a truly humongous mirror array, producing the necessary light levels (even at Outer System distances) shouldn't be a real problem.
Pressure compartments. This would seem to be a good idea, as Thucydides has pointed out, but the problem of light transmission remains. The more the compartmentalization, the more windows you need, and the weaker the structure becomes.
Population. It's probably a good idea to assume that a colony should be self-sufficient in terms of food, since food production is essentially a zero-sum game over the total community of colonies. So how much area do you need to feed people? Let's assume a semi (but not completely) vegetarian lifestyle, and go with 1 acre per person. I'd rather use 2 acres, but let's say the long daylight cycle will increase productivity. Meat is rabbits, fish and chickens, but no cows. 1 acre is roughly 4000 m², so the total internal area will support about 7500 people. That's a pretty small town, so something like representative democracy ought to work. Note that you can't get around this by assuming multi-level farms: it's light levels that are the limiting factor. Also to be considered is the need to recycle water and extract nitrates and phosphates. Since the system is a closed one, you can't keep adding fertilizer to keep up crop yields without poisoning the ecosystem. Also to be considered is the capital cost of soil. 30 million m² of dirt 1 foot deep will total about 10 km³. Nickel-iron asteroids have about 30% impurities such as silicates that conceivably could provide the basis for rock/sand/dirt, but it seems to take a lot of hand-waving to explain exactly how that would be separated from what was originally a molten blob. Again, I'm wondering about cost.
Space distribution. An obvious approach would be to build living quarters up the sides of the end caps. With a total cap area of about 1.5 million square meters, that's about 200 square meters per person. Note that that's not floor space, but window space. If the living space extends 100 meters along the axis at both ends, you only lose 200 meters out of 10 km, or 5% of your farmland, and total living space is about 20,000 m³ meters per person. Even allowing for common space and infrastructure, it seems adequate. It's a tossup as to which area would be more desirable — up towards the axis or down towards the surface. The natural first tendency to go for the axis has two things going for it — exclusivity (there's less area available) and low g luxury. The less-obvious drawback to this is that living at low g's is bad for one's health. The next obvious approach is the make the cylinder walls in multiple layers, with agriculture on the "roof", and industry "underneath". Since the materials involved are presumably nickel-steel, this ought to be straightforward, but rust due to groundwater might be a real long-term structural integrity problem. Industrial areas can obviously use artificial light.
Self-suffiency. As I mentioned in the farming section, I don't think that specialized "farm-habitats" make a lot of sense. In addition to the basic question of light levels needed, transportation would be a problem. Assume a habitat has specialized in something (industry, let's say) and has a population of 100,000, which isn't much by terrestrial standards. It will need something like 5 pounds of supplies per person per day, or 250 tons per day. Moving that sort of tonnage by spaceship is a bit iffy. Half of it (food) is fairly perishable, which means that transport needs a fairly high average speed over relatively low distances, which in turn implies high thrust. Worse, any technology has to be reactantless, since reaction mass is lost and has to be imported, and this is not likely to be developed Real Soon Now. (Photon drives are reactantless, it's true. At high thrust they make really excellent death rays, and traffic control gets very touchy.)
Specialization. Although each colony would be close to self-sufficient, there are intellectual economies of scale in technological activities, which might or might not be offset by ready communication. (Some companies have found, for instance, that outsourcing creates as many problems as it solves. Face-to-face meetings can be a very good thing.) This could lead to specialization among colonies, but would always be constrained by the cost of transportation of finished goods. Particularly, if local fabrication becomes simple (think of 3D printers on steroids), colonies might well become niche designers, with designs being sold and distributed rather than finished products or parts. Dissemination of artwork would be simple enough for writers and visual artists, and craft art (pottery, sculpture, paintings, custom artifacts, etc) would probably be cheap enough to ship to allow artist colonies to be established, although the need to provide food and basic services would keep the artistic population in check. Also, it might become standard in viable colonies (that don't tear themselves apart with internal conflicts) to require everyone to put in a certain amount of agricultural labor. To do otherwise would encourage a social split between the farmers and the (artists/engineers/bureaucrats/etc) which would have the potential to end very badly.