It would not so much be a sun 'gun' (i.e. not a single 'thing', not fast), but it would be able to destroy pretty much anything in orbit... unless that object was built 'sun-gun-bulletproof' :)
The suns radiation is filtered by the atmosphere - about half of the energy actually arrives on ground level. This filtering is not uniform, though - some wavelengths are completely filtered out, some come through pretty unhindered. For the reverse journey we have to account for this.
Both the orbit of an object and the suns angle of incidence at any given time are computable within incredibly tight margins.
The angle of a mirror to reflect the sunlight towards this object is thus also easily computable in advance. This makes the servos mechanically very unproblematic, we can build as many flat plate mirrors (for simplicitys sake; we could also go parabolic, but why complicate things for no real gain?) scattered over earths surface as we are willing to build. About 50% of our reflected light will actually make it through to space, now it's just a question of scale.
Reflectivity comes into play, but we can only build ~96% reflective, so the difference between targeting a 'perfect' reflecting body vs a black body is just that we need 20x more mirrors for the same effect.
If the target knows what's coming though, the going gets tough. Anything we heat up, itself starts to radiate, with a factor of T⁴; meaning for every doubling (2x) of temperature the object radiates (2x 2x 2x 2x) 16 times the energy it did at lower temperature. We need to overmatch that radiated energy to keep heating. An object around the melting point of tungsten (3700K) radiates 10627180W/m², i.e 10.5MW/m² - any other means of cooling vanish behind that.
So we need to get 11MW/m² up into LEO to make an object with a tungsten heat shield suffer. We need about 10m² of earthbound mirror to get 1kW/m² back into LEO, so 11 000m² of mirror array will do the trick. Let's go for overkill and have one square kilometer (1000 000m²) of individually aimed mirror plates, and we are golden.
This is not in the realm of a few IKEA mirrors and an Arduino, but we have both proven to have the aiming capabilities (telescopes solve the same problem), the capabilities in manufacturing precise mirrors (especially if we allow them to be flat, and go for an array instead of a perfect bowl) and the computational chops (again, telescopes) do solve this. The invest would be far, far beyond what a rocket with warhead would cost but its definitely feasible.