[llvm] [AMDGPU][NFC] Add tests in preparation for i8 vectorization (PR #138801)

[Jeffrey Byrnes via llvm-commits]

================
@@ -0,0 +1,1261 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
----------------
jrbyrnes wrote:

The name "vectorize-i8-as-i32.ll" is misleading I think -- can you rename to "vectorize-i8.ll"

https://github.com/llvm/llvm-project/pull/138801