[llvm] [VectorCombine] Fold bitcast(bitreverse.v8i8(bitcast(IntTy))) into bswap+bitreverse (PR #209037)

[Björn Pettersson via llvm-commits]

bjope wrote:

If the cost is equal for the vector and scalar version it seems like the IR isn't modified, right?

I wonder if we want to canonicalize to either form in such situations?

https://github.com/llvm/llvm-project/pull/209037