[llvm] [X86] Fold chained i32 CLMUL operands (PR #202250)

[via llvm-commits]

mygitljf wrote:

Thanks, I split this into three commits: test-only base codegen first, then the generic SelectionDAG change, and finally the X86 per-lane change. 

https://github.com/llvm/llvm-project/pull/202250