[llvm] [LV] Consider interleaving & tail-folding when -enable-wide-lane-mask=true (PR #163387)

[Kerry McLaughlin via llvm-commits]

================
@@ -4518,7 +4531,7 @@ LoopVectorizationPlanner::selectInterleaveCount(VPlan &Plan, ElementCount VF,
   // 3. We don't interleave if we think that we will spill registers to memory
   // due to the increased register pressure.
 
-  if (!CM.isScalarEpilogueAllowed())
+  if (!CM.isScalarEpilogueAllowed() && !CM.useWideActiveLaneMask())
----------------
kmclaughlin-arm wrote:

Adding a test with the optsize attribute showed that interleaving was still being considered. I’ve rewritten this to ensure we only consider interleaving when both `ScalarEpilogueStatus` indicates using predication and the wide lane masks flag is set.

https://github.com/llvm/llvm-project/pull/163387