[llvm] Reapply "[MemProf] Reduce cloning overhead by sharing nodes when possible" (#102932) with fixes (PR #106623)

[Teresa Johnson via llvm-commits]

================
@@ -2146,6 +2204,21 @@ bool ModuleCallsiteContextGraph::calleeMatchesFunc(
   return true;
 }
 
+bool ModuleCallsiteContextGraph::sameCallee(Instruction *Call1,
+                                            Instruction *Call2) {
+  auto *CB1 = dyn_cast<CallBase>(Call1);
----------------
teresajohnson wrote:

changed to cast here.

https://github.com/llvm/llvm-project/pull/106623