[llvm] [AMDGPU] Improve codegen for uniform f16<-->i32 conversions (PR #176833)

[Jay Foad via llvm-commits]

jayfoad wrote:

> > Why are f16<-->i32 conversions legal in the first place? I don't think any subtarget has an instruction for this, does it?
>
> I think no subtarget has it.

Raised #177342 so it doesn't get forgotten.

https://github.com/llvm/llvm-project/pull/176833