[llvm] [VectorCombine] Fold binary op of reductions. (PR #121567)

[Mikhail Gudim via llvm-commits]

mgudim wrote:

> still need to add cost driven tests

Added test for ARM. @RKSimon do you think more tests are necessary? You mentioned adding a test for X86 but, X86 doesn't implement `getMulAccReductionCost`. Can you clarify what test you had in mind

https://github.com/llvm/llvm-project/pull/121567