[llvm] [LLVM] treat `@llvm.ssub.sat` the same as `@llvm.aarch64.neon.sqsub` (PR #140454)

[via llvm-commits]

================
@@ -9861,14 +9843,15 @@ multiclass SIMDIndexedLongSD<bit U, bits<4> opc, string asm,
 }
 
 multiclass SIMDIndexedLongSQDMLXSDTied<bit U, bits<4> opc, string asm,
-                                       SDPatternOperator Accum> {
+                                       SDPatternOperator VecAcc,
----------------
SpencerAbson wrote:

Perhaps we could leave a comment here (and something in the commit message) about why we still keep `int_aarch64_neon_uqadd` around for scalar types (https://reviews.llvm.org/D69374), it looks a little strange without context.

I couldn't think of a nice way to split these patterns at the time, I'll read it again tomorrow. 

https://github.com/llvm/llvm-project/pull/140454