[llvm] [CodeGen] Avoid sinking vector comparisons during CodeGenPrepare (PR #113158)

[David Green via llvm-commits]

davemgreen wrote:

Yeah thanks. A lot of the tests look like expanded masked load/store, which are going to be a bit different to more normal codegen due to the scalarization. At least in some cases it might be better to scalarize the compare. Hopefully they shouldn't come up much in practice though.

https://github.com/llvm/llvm-project/pull/113158