[llvm] [VectorCombine] Scalarize bin ops and cmps with two splatted operands (PR #137786)

[Luke Lau via llvm-commits]

lukel97 wrote:

Thanks for the reviews, I've opened up #137948 to fix this from the InstCombine side.

The codegen diff on RISC-V from that PR is actually a lot more minor than I expected, and I didn't notice any missed splat patterns. The only thing is that it also needs https://github.com/llvm/llvm-project/pull/137823 to make sure that everything still gets scalarized. 

Otherwise it subsumes the improvements seen from this PR, so I'll close this now. 

https://github.com/llvm/llvm-project/pull/137786