[compiler-rt] [compiler-rt][ARM] Optimized double-precision FP mul/div (PR #179923)

[Simon Tatham via llvm-commits]

statham-arm wrote:

This is the current head of my patch queue of optimized FP implementations. I've just merged out from main to pick up the landed version of the previous patch, and updated the `#include`s.

After this one, the next two patches in the queue are the already-approved float and double comparisons (which will need last-minute rebasing and the same `#include` fix, but nothing more major).

https://github.com/llvm/llvm-project/pull/179923