Pretty tremendous stuff here.
I have to say that the perception/action model you drew up is just stupendous.
The thing that I find really interesting about it is the role of gestalt understanding vs. more "modular" understanding. I think that's one reason why very often talented individuals are poor coaches/teachers. They rapidly develop a gestalt understanding that moves into the realm of automation and intuition very quickly, and they are then unable to verbalize that understanding, because their model was never developed to same extent as someone who had to wrestle with the movement or material.
This is one reason I like teaching (and arguing), as it forces me to develop models more fully than I might otherwise. So even though we seek to have a visual gestalt of the squat so that we can see instantly what someone is doing wrong, we also need a thorough model that can be transmitted through the various modes of communication to the lifter, to help refine their internal model of the squat (which is what we're really doing, to some extent).