Firstly - this is totally
brilliant. My reaction was "WOAH" *think* *think* "...COOOOL!"
Then I tested the theory.
I firstly fired up the Thrustmaster analyzer, and tested my corners. I dont think I'd used my warthog since the last reboot.
Tried top left corner - best I could do was 95,95. I could get to 0,y and x,0 - just not at the same time - so far, so good.
Then I tried to get other corners - and got the same behavior - I was initially about 64500,64500 in the bottom right corner..
Then I applied just a little more force to see what I could really get the corners too. And I could get to 0,0, and 65535,65535. And now its "easy". There seems to be no edge deadzone, I can get to the full corners, and it seems to "work" like a square bounding box.
So my guess is that there is some self-calibrating going on in the first instance, and I wonder if there is some fancy maths going on in the firmware too.
So - at least on my stick - it seems to be maybe not really an issue.
I really admire the fact you came up with the idea and then coded it. When I have some more time I'm certainly going to dig into your code to examine the method of what your doing. It will certainly be useful to me and others somewhere