accessSubscriptsAreUnrolledLoops: temporarily remove type deduction

This allows TensorReferenceGroup.originalAccesses() to be converted
to templated isl types without having to convert
accessSubscriptsAreUnrolledLoops at the same time.
Automatic type deduction will be reintroduced
when accessSubscriptsAreUnrolledLoops is converted.

Author: Sven Verdoolaege

tc/core/polyhedral/cuda/memory_promotion_heuristic.cc

@@ -341,7 +341,7 @@ bool accessSubscriptsAreUnrolledLoops(
     unrolledDimsMupa =
         unrolledDimsMupa.intersect_domain(group.originalAccesses().domain());
 
-    auto accesses = group.originalAccesses();
+    isl::union_map accesses = group.originalAccesses();
     auto schedule = outerSchedule.flat_range_product(unrolledDimsMupa);
     accesses = accesses.apply_domain(isl::union_map::from(schedule));