[llvm] Reapply "[MemCpyOpt] support offset slices for performStackMoveOptzn and processMemCpy", with bug fixed (PR #180002)

[Jameson Nash via llvm-commits]

================
@@ -57,11 +57,13 @@ class MemCpyOptPass : public PassInfoMixin<MemCpyOptPass> {
 
   LLVM_ABI PreservedAnalyses run(Function &F, FunctionAnalysisManager &AM);
 
+  // Glue for the old PM.
+  LLVM_ABI bool runImpl(Function &F, TargetLibraryInfo *TLI, AAResults *AA,
+                        AssumptionCache *AC, DominatorTree *DT,
+                        PostDominatorTree *PDT, MemorySSA *MSSA);
----------------
vtjnash wrote:

As I've said before, the ids-check bot required this change be added to this PR (even though I agree ids-check possibly shouldn't be this aggressive, I cannot fix or ignore the CI bot in this PR): https://github.com/llvm/llvm-project/pull/176436#discussion_r2705543343

https://github.com/llvm/llvm-project/pull/180002