Oh, yeah, about the back surface needing to be perfect… I remember that the main reason is that the mirror is supported in the telescope by 160 actuators that can apply adjustable pressure to each mounting point, which allows the mirror to be held in the same exact curvature at which it is tested, even as it’s tilted in the telescope to be pointed up at various angles.
Each actuator presses against a triangular cradle that presses against three points on the mirror itself. These mounting points are glued onto the rear of the mirror. They need to be just ever so, to ensure that the mirror is supported within a few tens of nanometers of its required flatness.
