Spent the whole day on OpenGLContext.  I bashed out a "molehill" tutorial based on the classic from Mark Kilgard), mostly to verify that the NURBS stuff really isn't broken[1], and did a bit of work creating a "gear" geometry type (not satisfied with that one at all, very dull looking).  I made the ROUTE mechanism far easier to use (so it's actually convenient to use ROUTES instead of manually creating timers), and fixed a bug in ROUTE/IS registration/usage by the OpenGLContext module.  I also fixed a few tests/tutorials to use scenegraphs instead of custom rendering code. I converted two other tests into tutorial format along the way, though I haven't yet got a good handle on the tutorial path for the scenegraph-based stuff.

Need to either re-patch pydoc to regenerate the overall PyOpenGL/OpenGLContext/PyVRML97 pydoc set or switch to something else to generate the files.  The patched pydoc no longer recognizes the OpenGL entry points, as they are instances, rather than functions/built-ins.  Also need to figure out why, on Win32 with PygameContext a resize of the window seems to be losing all textures (almost as if the GL context is being recreated rather than just resized).  No problem on Linux.

[1] NURBS seem to really misbehave when rendered into a display list these days.  My working theory is that the evaluator is trying to do something fancy with e.g. object-space-tess or window-space tess and the display-list compile environment doesn't match that assumption.