[llvm] [AArch64] Don't allow mixed partial reductions without i8mm (PR #137602)

[Sam Tebbs via llvm-commits]

================
@@ -225,8 +225,203 @@ for.exit:                        ; preds = %for.body
   ret i32 %add
 }
 
+define i32 @dotp_z_s_neon(ptr %a, ptr %b) #1 {
----------------
SamTebbs33 wrote:

Done. I've called this one sudot_neon since it's `mul(sext, zext)`.

https://github.com/llvm/llvm-project/pull/137602