[llvm] [AMDGPU] Fold saddsat(x, -C) to ssubsat(x, C) when C is an inline constant (PR #216592)

[Madhur Kumar via llvm-commits]

MadhurKumar004 wrote:

Yes it can be extended to i16, but it wont work on unsigned cases because the unsigned operand cannot be negative - it will be treated as a large positive value.

https://github.com/llvm/llvm-project/pull/216592