Float16

Already filed FB7747586 for this, but I’d be interested to know if there’s a good way to use simd functions on half-precision (that is, __fp16 or Float16) types. I have a codebase that I maintain in both C and Metal, and although half is supported on the Metal side, and although Swift now has a Float16 type, the simd framework functions don’t know anything about half-precision and so code that works on one can’t be easily ported.