MJPEG Tools Svn

Fix a couple typos in comments. Tidy up some usage info messages.

Tidy up a little. There are functions that can be used to generalize the
computation of the chroma buffer sizes.

It is not the cause of the crashing on a PPC but the use of free() was
totally broken. free() takes the pointer to the malloc'd memory, you should
not subtract something from the pointer first!

For some time now there's been the ability to selectively disable accelerated
functions (mjpeg_simd_helper gives a list of the functions for which this can
be done). Alas, that only worked for the X86 (MMX, SSE) functions. The
changes being checked in now enable the same capability for the Altivec routines

This is being done to work around an apparent limitation in the sad_00 routine.
yuvdeinterlace wants to use sad_00 with data aligned on 8 byte boundaries.
This works for MMX, SSE but not Altivec. So now with the simd_disable()
implemented it will be possible to disable sad_00_altivec within yuvdeinterlace.

fixed bad pointer on free()

Fix typo in usage message.

mjpegtools does not (and has not for a long time) depended on or used nasm.

jpeg-mmx is dead and gone.

.

Andrew indicated that the seqstats.hh and seqstats.cc are "not quite orphaned".
BUT, seqstats.cc is used NOWHERE. seqstats.hh is only used in seqstats.cc.
Neither compile. So they shouldn't be deleted. Put them in an out of the
way place like EXTRA_DIST so they remain in the .tar.gz.