promoteToSharedBelow: take into account the mapping to blocks

Currenly, promotion to shared memory is only performed below the loops
mapped to blocks.  Thus, tensor reference groups implictly account for
blocks.  The scope of mapping in the tree will be changed in upcoming
commits.  Explicitly include block mapping into the partial schedule
within which the tensor reference groups are computed.

Author: ftynse

tc/core/polyhedral/cuda/memory_promotion_heuristic.cc

@@ -440,9 +440,10 @@ void promoteToSharedBelow(
     size_t& remainingMemory) {
   auto root = scop.scheduleRoot();
   auto partialSched = partialSchedule(root, bandNode);
+  auto mapping = collectMappingsTo<mapping::BlockId>(scop);
 
   auto groupMap = TensorReferenceGroup::accessedWithin(
-      partialSched, scop.reads, scop.writes);
+      partialSched.intersect_domain(mapping), scop.reads, scop.writes);
   // Pure affine schedule without (mapping) filters.
   auto partialSchedMupa = partialScheduleMupa(root, bandNode);