Replaced by fabs() that seem to work ok

Found by LeakSanitizer in #494.

jar_xm.h does some shady pointer casts leading to unaligned accesses
and breaking strict aliasing. x86 has special circuitry for doing
unaligned accesses, but on other architectures, it may trap and require
kernel fix-up or crash outright. With this patch, one obstacle in
porting raylib to the GameBoy Advance has been removed. Go for it ;-)

To avoid having to rewrite that `mempool' code, insert padding before
structs and instruct the compiler (GCC, most importantly), to be gentle
when optimizing.

This fixes #490 (Unless we got ourselves 256-bit pointers, if so,
hello future!)

Problems when compiling with VS2015
Probably DEBUG(...) macro is not the best naming...

Still some work left...

Moved all external libraries used by raylib to external folder inside
raylib src. Makefile has already been update and also the different
includes in raylib modules.

Boolean errors

This is an early draft, needs lots of work. Still need to figure out way
to calculate total length of song. This is hard because xm tracks stream
out zeros when done, only position in track can be found. Position does
not give any direct value of how much more time is left. I think that by
setting the loop count to 1 and seeking until the end I can total up the
number of samples and come up with a length.