[AD] /src/i386/iblit32.s MMX alignment

[aj <aj@xxxxxxxxxx>]

from my limited understanding of ASM, the  /src/iblit32.s   MMX/SSE code 
opts out to a non-mmx implementation if the bitmap->w  is not 4 pixel aligned.
the code is fine, but the docs make no mention of it.

Maybe a line in the docs near blit() or masked_blit()  could mention 4 
pixel alignment will typically be faster.

aj.