[llvm] [X86] Prefer `lock or` over mfence (PR #106555)

[Phoebe Wang via llvm-commits]

================
@@ -754,6 +754,10 @@ def TuningUseGLMDivSqrtCosts
 def TuningBranchHint: SubtargetFeature<"branch-hint", "HasBranchHint", "true",
                                         "Target has branch hint feature">;
 
+def TuningAvoidMFENCE
----------------
phoebewang wrote:

Should better to use the opposite definition, e.g., `TuningPreferMFENCE` given modern CPUs don't need it?

https://github.com/llvm/llvm-project/pull/106555