[llvm] [LoongArch] Enable the TypePromotion pass from AArch64 (PR #98868)

[Lu Weining via llvm-commits]

================
@@ -6,12 +6,14 @@ define i1 @andn_icmp_eq_i8(i8 signext %a, i8 signext %b) nounwind {
 ; LA32-LABEL: andn_icmp_eq_i8:
 ; LA32:       # %bb.0:
 ; LA32-NEXT:    andn $a0, $a1, $a0
+; LA32-NEXT:    andi $a0, $a0, 255
----------------
SixWeining wrote:

I suppose this regression will be addressed by later patch.

https://github.com/llvm/llvm-project/pull/98868