[clang] [llvm] [mlir] [Flang][OpenMP] Enable no-loop kernels (PR #155818)

[Michael Kruse via llvm-commits]

================
@@ -0,0 +1,96 @@
+! REQUIRES: flang, amdgpu
----------------
Meinersbur wrote:

```suggestion
! REQUIRES: flang
```
Shouldn't this still be correct with nvidia GPUs?

https://github.com/llvm/llvm-project/pull/155818