Optimizing Vector Math for Debug Compilation

I changed the title a bit from Aras’s original because I felt that the original title wasn’t as clear. Hopefully that’s okay with folks.

Aras does a lot of great deep dives into optimizing graphics programs and this one focuses on optimizing the performance of Debug builds for Blender.

I am pleasantly surprised how well Clang performs here, but conversely shocked at how poor MSVC is doing.