[llvm] [NVPTX] Generalize and extend upsizing when lowering 8/16-bit-element vector loads/stores (PR #119622)

[Artem Belevich via llvm-commits]

================
@@ -56,6 +56,42 @@ define half @ld_global_v8f16(ptr addrspace(1) %ptr) {
   ret half %sum
 }
 
+; CHECK-LABEL: @ld_global_v8i8
----------------
Artem-B wrote:

We should convert this test to auto-generated checks, too.

https://github.com/llvm/llvm-project/pull/119622