impl/gcc: add native 256-bit and 512-bit intrinsics these are simple to implement. At some point I'd like to refactor vec into using a union and being able to detect AVX512 and friends at compile time, so that the processors that *can* use it are enabled at runtime. This would mean adding a vec_init function, which isn't that big of a deal and can just be run at startup anyway and will grab the CPU flags we need.