OK good explanation. To continue the purely academic discussion of unimportant technical matters, my thoughts are that what you are seeing at high alpha is increased adverse yaw combined with increased yaw/roll coupling from high dihedral in ag planes. That's the airplane you're talking about right? It makes sense that roll response would be slightly delayed by not leading with the rudder, since leading with rudder should produce a nearly immediate roll response due to dihedral effect. But IMO, that doesn't necessarily mean this rudder-lead turn is perfectly free of momentary skid. Again, the skid ball may not necessarily show a skid. If the ball is mounted above the roll axis, the ball will naturally displace slightly in the slipped direction while rolling into a perfectly coordinated turn. It's pure momentum, just like when your upper body gets pushed in the opposite direction of the roll, assuming your upper body is above the roll axis.
So if the skid ball in the airplane is prone to this effect, a slightly skidded turn initiation could actually cause the ball to stay perfectly centered, since the skid is opposing the opposite reaction of the ball. And if humans in general seem to have slow feet, I think this only worsens in high AOA conditions, making it seem like the lead w/ rudder requirement exists, or is made more obvious. Good discussion, but I'm still not convinced that a truly properly timed, and sufficiently deflected rudder input cannot immediately control adverse yaw and maintain perfect coordination. But maybe that assumes a computer is flying the airplane. Humans just aren't necessarily that precise, and I guess do what they need to do to get good results. Nobody is saying you or other 'rudder leaders' don't get decent results, or can't fly an airplane.