lcms2-2.20 is a maintenance release
Changes:
- Added a plug-in type for profile header parsing.
- Fixed premultiplied alpha on planar 16 bit and optimized matrix-shaper transforms.
- Fixed planar half float and Lab double formatters, and planar strides in fast_float.
- Fixed gamut checking with two profiles and a possible crash with malformed profiles.
- Recovered neutrality correction in black point detection.
- Made cmsCreateLab2Profile white point handling consistent with cmsCreateLab4Profile.
- Fixed UTF8 buffer size calculation.
- CGATS parser is now more permissive on certain properties.
- Fixed callback type mismatch in resampled CLUT evaluation.
- Fixed a crash in PostScript CSA generation.
- Fixed memory leaks in black preserving intents and .cube loading.
- cmsChangeBuffersFormat now correctly rejects format changes on floating point transforms.