[AArch64][GlobalISel] Legalize narrow scalar FP arithmetic

Widen narrow fp arithmetic ops (e.g. G_FADD). When we don't have full FP16
support, widen to s32. Otherwise widen to s16.

https://godbolt.org/z/TbT9Pqa7e

Differential Revision: https://reviews.llvm.org/D108660