[llvm] [VectorCombine] Fold concat(binop(a,c), binop(b,d)) -> binop(concat(a,b), concat(c,d)) for i1 vectors (PR #206087)

[Aayush Shrivastava via llvm-commits]

================
@@ -0,0 +1,49 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
----------------
iamaayushrivastava wrote:

Thanks for catching that! You're right, `llc` doesn't run VectorCombine, so the test wasn't exercising the change at all. I've replaced it with a proper `opt -passes=vector-combine` test under `Transforms/VectorCombine/X86/shuffle-of-binops-i1.ll` that directly verifies the fold fires (and is correctly suppressed for multi-use binops).

https://github.com/llvm/llvm-project/pull/206087