Chris,
OK - that gives me some idea - I would guess that the machine marks would be at least +/- .001", maybe more.
We should consider what texturing does to a beam, and maybe just create a small "imperfection" in a reflector, such as a very small dent or punch mark, and see to what extent it affects the reflected light.
I'm also thinking that "where" a surface variation appears on a reflector might have a greater or lesser affect (or produce a different type of effect), depending on whether it is close to the filament vs farther out near the rim, etc.
I wish I had more junk reflectors to play with. I'm a "try it and see" kind of guy when I have the need to quickly establish a starting point based on cause/effect relationship experiments. I call this my "shotgun" method of doing debug on custom automated machinery. It works very well for me, and has taught me a great many things.
Later - John