[llvm] [X86] Add vXi64 MULHU lowering, keeping full-width products scalar (PR #206983)

[Rito Takeuchi via llvm-commits]

Licht-T wrote:

@phoebewang @Andarwinux I meant the performance regression between pre-#169819 and #169819, not between microarchitectures. The key is #169819 degrades the div performance by 0.48x on SPR w/ AVX-512.

Please see Result 3 in https://github.com/llvm/llvm-project/pull/169819#issuecomment-4845099097 for more details.

https://github.com/llvm/llvm-project/pull/206983