You're kidding, right?
If not, then you should know that reflectors are more or less parabolic. The "more or less" part is known as the divergence and is measured in degrees. Most reflectors have some divergence. The less divergence, the tighter the spot (theoretically). The problem is that if there is no divergence, and you are using a non-point source (which you are), then the beam will only be a spot at a certain distance. Mag solved this problem by having a cam action focus. Personally, I prefer a reflector with a decent amount of divergence in it, like a SureFire, and OP coating, that yields a beam good at a large range of distances.
Moving on, the tighter the parabola, and the more of it you have, then the more it is going to throw and be a spot beam. The opposite of these gives you a flood. Also, the more you defocus the filament from the focal point, the more you will have a flood, plus a "hole" in the beam, of course.
As for shiny and smooth, that comes with a good machining job, plus undercoating, plus vacuum metalizing. Usually this is done with aluminum, but there are other choices. There are comanies who could do this for you. NewBie posted a thread on this at one point.
As for the "main question" of how light "comes out" of the LED or incan bulb, that's a real corker, that is. It does indeed "come out". In an incan filament which is wound around a round core, the radiation pattern is largely cylindrical, around the filament in the way you'd expect. If the filament is transverse, it is oriented perpendicular to the axis of the reflector bore. If the filament is axial it is oriented parallel and coincident with the reflector bore.
So, tell me, how do you plan on machining your reflectors? Do you have some kind of sophisticated setup, like a Swiss multi-axis CnC machine or something?
You know that companies like Carley Lamps sell reflectors of various diameters, and coatings, right? Wouldn't it be easier and better to buy your own?