BFS optimization

• Add pass to optimize stores in BFS.

juno_samples/rodinia/bfs/src/gpu.sch

@@ -39,6 +39,7 @@ fixpoint {
 simpl!(collect);
 
 fork-tile[1024, 0, false, true](init, traverse, collect);
-fork-split(init, traverse, collect);
+let out = fork-split(init, traverse, collect);
+simpl!(*);
 
 gcm(*);