fix incorrectly-named assert, attempt to fix (char) cast warning (tricky since char's signedness is unknown)

this variable was causing a warning while compiling with gcc.

tweak docs

Also ran a bunch of test cases to make sure the IDCT and H2V2
resamplers were correct.

Put progressive JPEG AC decode logic back the way I wrote it originally (I changed it to match jpgd when I was trying to figure out why it didn't work);
add STBI__ prefixes to internal SCAN_ enum;
strip unused function arguments for progressive funcs;
tweak release notes;
forget to git commit frequently so these would all be in their own commits;

stbi__ prefix on two functions that were missing it

fix long-standing bug where gifs were just broken AFAICT

tweak credits
tweak revision history
remove trailing whitespace

fix a few old error messages

some documentation cleanup

add missing STBIDEFs to a few functions;
update documentation with full 1.49 info even though most isn't implemented yet

  support using inline asm for cpuid

YCbCr:
  switch SSE code to constants that match old C;
  create C version that is same as SSE;
  tiny optimization(?) of SSE

Also add SSE2 detection for MSVC++. Detection on GCC will follow
later.

I want to support SSE2 kernels that auto-detect. If implemented
using globals, this would trigger thread race checkers if stb_image
was used in multiple threads. Just make the kernels be explicitly
per-stbi__jpeg to avoid this.

(This commit is the first step in replacing existing STBI_SIMD
functionality.)

Inside decode_block, we're still sparse, and we can use that
sparsity pattern without doing extra work to discover it.