[llvm] [VectorCombine] Support simplification to scalar store for multiple insertelt (PR #132820)

[Simon Pilgrim via llvm-commits]

================
@@ -1962,62 +1962,138 @@ static Align computeAlignmentAfterScalarization(Align VectorAlignment,
 //   %0 = bitcast <4 x i32>* %a to i32*
 //   %1 = getelementptr inbounds i32, i32* %0, i64 0, i64 1
 //   store i32 %b, i32* %1
-bool VectorCombine::foldSingleElementStore(Instruction &I) {
+bool VectorCombine::foldInsertElementsStore(Instruction &I) {
----------------
RKSimon wrote:

foldInsertElementStores ? It create multiple stores.

https://github.com/llvm/llvm-project/pull/132820