[VectorCombine] Fold binop(shuffle(V1, Mask), shuffle(V2, Mask)) -> shuffle(binop(V1, V2), Mask)

InstCombine currently folds this here:
https://github.com/llvm/llvm-project/blob/acfaabebcfd10d4dbabeb54cf0c60509648fb82f/llvm/lib/Transforms/InstCombine/InstructionCombining.cpp#L2590-L2605

However, we're missing an equivalent, cost driven fold in VectorCombine, which prevents some later folds occurring in the pass so we have to wait for later InstCombine run to handle it.