Compile-time tuning

Specialized the dataflow iterators and did a few other minor tweaks.
Showing ~5% compile-time improvement in a single-threaded environment;
less in multi-threaded (presumably because we're blocked by something
else).

Change-Id: I2e2ed58d881414b9fc97e04cd0623e188259afd2
diff --git a/compiler/dex/mir_analysis.cc b/compiler/dex/mir_analysis.cc
index d7a4136..8472a3c 100644
--- a/compiler/dex/mir_analysis.cc
+++ b/compiler/dex/mir_analysis.cc
@@ -1061,7 +1061,7 @@
   memset(&stats, 0, sizeof(stats));
 
   ClearAllVisitedFlags();
-  AllNodesIterator iter(this, false /* not iterative */);
+  AllNodesIterator iter(this);
   for (BasicBlock* bb = iter.Next(); bb != NULL; bb = iter.Next()) {
     AnalyzeBlock(bb, &stats);
   }