2.0

· Greatly improved compilation times by removing most SFINAE
· Added NEON64 support
· Added scalar version of library for platforms without hardware vectors
· Added float8 type, backed by AVX
· Improved float4x4 matrix by using AVX vectors when available
· Added doubleN types. Both SSE and AVX
· Added uintN types
· Added missing hlsl functions like fmod, mad, sincos