[clang] [CIR] Use declarative TableGen constraints for overflow flag verification (PR #195476)

[Henrich Lauko via cfe-commits]

================
@@ -134,7 +134,7 @@ void test_sized_array_delete(SizedArrayDelete *ptr) {
 // LLVM: [[DELETE_NOTNULL]]:
 // LLVM:   %[[ALLOC_PTR:.*]] = getelementptr i8, ptr %[[PTR]], i64 -8
 // LLVM:   %[[NUM_ELEM:.*]] = load i64, ptr %[[ALLOC_PTR]], align 4
-// LLVM:   %[[ARRAY_SIZE:.*]] = mul i64 4, %[[NUM_ELEM]]
+// LLVM:   %[[ARRAY_SIZE:.*]] = mul i64 %[[NUM_ELEM]], 4
----------------
xlauko wrote:

Yeah, that is unfortunate property of MLIR canonicalizer putting all constants to "suffix" on commutative operations. :/ 

https://github.com/llvm/llvm-project/pull/195476