Fixed a couple of bugs in the general and X11 shape code, and fixed a bug in testshape that was keeping it from recognizing surfaces without alpha.  Thanks to Andreas's bit-bashing tip, X11 shaped windows now work entirely, AFAICT.

Wrote out the system for breaking shape-masks into quad-trees of rectangles, and added code to conglomerate those quad-trees of rectangles into regions for setting shapes under Win32.

Rejiggering the way shaped windows are created as preparation for OS X implementation.  Fixed overdrive bug in test program that appears to have been introduced by someone other than myself.

Reapplied Win32 make-it-build patch.  What's going on?  Still get rendering artifacts when testing on x11, have literally no idea why.

Put the render loop back in its right place, but with delaying functionality to keep it down to roughly 60fps, not eating up all the X11 time.

Boxed up the rendering loop and turned it into a redraw/update function.  This makes things way more responsive, usable again.

Minor bugfixes.  testshape now draws a shaped window with bizarre, pixellated gashes of transparency across it, and in doing so seems to hog a system resource and slow the rest of the video system down.

Same place as before, but optimizing a bit to try to isolate the spot in the program that locks things up.

Rewrote test program for shaped windows.  It definitely displays recognizable pictures now, but the resizing and shaping functionality isn't behaving correctly, possibly due to a miscalculation of alpha values.

Updated testeyes.c.  It now shows some kind of gibberish Missingno-type picture somewhere on the screen and, possibly, hangs the window system.  On the upside, the Missingno is definitely a non-rectangular window.