Here's a shot at updating the dynamic X11 code.

- It no longer needs all code to prefix a 'p' before X calls...this makes development more natural, and makes it easy to totally remove the dynamic code if this doesn't work out. The bulk of this patch is reverting all the 'p' prefixing in the rest of the source code.
- It now keeps track of what did and did not load by "modules", so the build system may have support for, say, the UTF-8 entry points, but these can be missing on the end-user system...the basic Xlib entry points will still work in this case, but the code can check SDL_X11_HAVE_UTF8 to see if it's safe to call the functions in this module. For SDL builds linked directly against Xlib, these module checks are always true.
- Current moudles are: "BASEXLIB" for all the general, must-have functionality...the X11 target will gracefully fail to initialize if any of those symbols are missing. "UTF8" for the Unicode stuff and XIM support (we gracefully fallback to single-byte char handling if this is missing). "SHM" for the MIT Shared Memory extension support (we do without if it's not there). There are some other dummy modules for internal symbols that aren't called if they don't exist on a system, like the 32-bit IO fallbacks on amd64, etc.
- Future modules will be "XRANDR", and whatever else we don't want to link against directly. We could possibly jettison some of the included code for things like xme and Xinerama and such, too.

Let me know what you think.

--ryan.