lcms2-2.12-150400.1.10

- update to 2.12:
  * Added build system for fast-float plugin (see plugin documentation)
  * Added new build-in sigmoidal tone curve
  * Added XCode 12 project
  * Added support for multichannel input up to 15 channels
  * Fix LUT8 write matrix
  * Fix version mess on 10/11
  * Fix tools & samples xgetopt
  * Fix warnings on different function pointers
  * Fix matlab MEX compilation
  * plugin: cleanup and better SSE detection
  * plugin: add lab to any on float
  * plugin: it can now be compiled as C++
  * recover PDF documentation, but try to keep it under a resonable size.
  * Prevent a rare but possible out-of-bounds read in postscript generator
  * Fix some compiler warnings
  * Add named color profile building sample to testbed

- Update to 2.11
  * Fixed __cpuid() on fast float plugin to allow gnu gcc
  * Fixed copy alpha bounds check
  * Fixed data race condition on contexts pool
  * Fixed LUT16 write matrix on multichannel V2 profiles
- Changes in 2.10
  * Added a compilation toggle to remove "register" keyword in API.
  * Previously commercial, fast_float plug-in is now released as open source
    under GPL3 license.
  * MD5 functions are now accesible through plug-in API.
  * Added support for Visual Studio 2019
  * Bug fixing.
- Drop lcms2-cgats-memory-allocation.patch, its now in upstream code
- Refresh patches
- Run spec-cleaner

- Update to 2.9
- Drop 0001-Added-an-extra-check-to-MLU-bounds.patch, it is now
  in the upstream code
  * Bugfix release
  + Several fixes related with security, and therefore not detailed here.
  + C++ compiles now without warnings
  + Added OSX and clang in travis
  + Added a travis-ci test matrix for many compilers and OS.
    Thanks to Thomas Weber (debian) for this great improvement.
  + testbed can now work with dynamic versions of library
  + Fixed wrong planar formatters regarding linestride interpretation