Merry Christmas! 🎄
- Added ldexp
- Add a simple half class and f16tof32/f32tof16 functions
- Add uint and enforce int to be 4 bytes
- Added first collection of data packing functions
- Fix incorrect mapping of _hlslpp_subm_ps to vmlsq_f32 which was causing sin_ps and cos_ps to return incorrect values on ARM
- Enforce alignment on types to avoid platform mismatches
- Fix clang-cl + ARM64 platform combination