rework Photoneo YCoCg(-R) for SIMD autovectorization

[bmegli]

Related to #15, #13

Photoneo YCoCg(-R) subtly differs from VESA DSC YCoCg-R

• https://github.com/photoneo-3d/photoneo-cpp-examples/issues/4

  • extra check for y==0 zeroying out pixel to protect black pixels in 4:2:0 subsampling
  • potentially small differences in LSbs, some of which are already being corrected
  • when scaling to 8 bit RGB this difference is not visible anyway

• rework our implementation so that it autovectorizes

• add commented out CMake line to enable autovectoratization

  • do not enable native optimization with vectorization for generic build
  • code generated this way may not be portable to different architectures

---

Powersafe

• current implementation is the last column

No optimization	Optimized	Autovectorized without y==0 check	Autovectorized with y==0 check	Autovectorized with optimized y==0 check
up to 30 ms	up to 11 ms	2-3 ms	Up to 7-8 ms	2-3 ms

---

Performance

• current implementation is the last column
• for reference, the code would not typically run this way

No optimization	Optimized	Autovectorized without y==0 check	Autovectorized with y==0 check	Autovectorized with optimized y==0 check
7 ms	3 ms	1 ms	2 ms	1 ms