[llvm] [Offload] Add MPI Proxy Plugin (PR #114574)

[Jhonatan Cléto via llvm-commits]

cl3to wrote:

Here are some results comparing the performance of RSBench and XSBench (lookup_size=10^10) when executed using the MPI proxy plugin (MPP) and the CUDA plugin.

**RSBench Strong Scaling:**

![rsbench_strong_scaling_speedup](https://github.com/user-attachments/assets/e1865383-6242-49b2-a3cf-6d2036f94ead)

**XSBench Strong Scaling:**

![xsbench_strong_scaling_speedup](https://github.com/user-attachments/assets/32f7eb29-773b-4cc6-a59b-380af6dc73b9)

**Weak Scaling using MPP:**

![proxy_app_weak_scaling](https://github.com/user-attachments/assets/67cd76f7-9ac1-41fa-a6aa-8a29abb2862d)



https://github.com/llvm/llvm-project/pull/114574