[llvm] Add clmul zext AArch64 lowering tests (PR #179641)

[David Green via llvm-commits]

================
@@ -455,4 +455,797 @@ define <2 x i32> @clmul_v2i32_neon(<2 x i32> %x, <2 x i32> %y) {
 ;   ret <1 x i64> %a
 ; }
 
-attributes #0 = { "target-features"="+aes" }
\ No newline at end of file
+define <16 x i8> @clmul_v16i8_neon_zext(<16 x i4> %x, <16 x i4> %y) {
----------------
davemgreen wrote:

I would remove i4>i8 as i4 is a difficult type they are always going to be a bit odd. It might be worth having a v2i64->v2i128 test though, as AFAIU it might come up in crypto algorithms where they often use i128 types.

https://github.com/llvm/llvm-project/pull/179641