[llvm] [X86] Optimize (a & 0xff) | ((b & 0xff) << 8) into high-byte register insertions (PR #214645)

[Craig Topper via llvm-commits]

topperc wrote:

Please stop force pushing unless you are changing the tests in the first commit. You have been warned about this before.

https://github.com/llvm/llvm-project/pull/214645