[llvm] [LLVM][OpenMP] Allow OpenMPOpt to handle non-OpenMP target regions (PR #67075)

[Joseph Huber via llvm-commits]

jhuber6 wrote:

Is the desire to not run OpenMPOpt on kernels generated from CUDA? It would probably be much easier to make the CUDA metadata match OpenMP's otherwise.

Also unrelated complaint, I really wish we could unify the kernel annotations between AMDGPU and NVPTX.

https://github.com/llvm/llvm-project/pull/67075