[memprof] Use std::move in toMemProfRecord (#93133)

std::move and reserve here result in a measurable speed-up in
llvm-profdata modified to deserialize all MemProfRecords.  The cycle
count goes down by 7.1% while the instruction count goes down by 21%.

Author: kazutakahirata

llvm/lib/ProfileData/MemProf.cpp

@@ -247,13 +247,15 @@ MemProfRecord IndexedMemProfRecord::toMemProfRecord(
         Callback) const {
   MemProfRecord Record;
 
+  Record.AllocSites.reserve(AllocSites.size());
   for (const memprof::IndexedAllocationInfo &IndexedAI : AllocSites) {
     memprof::AllocationInfo AI;
     AI.Info = IndexedAI.Info;
     AI.CallStack = Callback(IndexedAI.CSId);
-    Record.AllocSites.push_back(AI);
+    Record.AllocSites.push_back(std::move(AI));
   }
 
+  Record.CallSites.reserve(CallSiteIds.size());
   for (memprof::CallStackId CSId : CallSiteIds)
     Record.CallSites.push_back(Callback(CSId));