Add fix for slow GPU aware MPI to JSC booster script

[AlexanderSinn]

In my testing so far, setting this environment variable greatly improves performance of GPU-aware MPI on the booster. 

export UCX_CUDA_COPY_REG_WHOLE_ALLOC=on

• [ ] Small enough (< few 100s of lines), otherwise it should probably be split into smaller PRs
• [ ] Tested (describe the tests in the PR description)
• [ ] Runs on GPU (basic: the code compiles and run well with the new module)
• [ ] Contains an automated test (checksum and/or comparison with theory)
• [ ] Documented: all elements (classes and their members, functions, namespaces, etc.) are documented
• [ ] Constified (All that can be const is const)
• [ ] Code is clean (no unwanted comments, )
• [ ] Style and code conventions are respected at the bottom of https://github.com/Hi-PACE/hipace
• [ ] Proper label and GitHub project, if applicable

[MaxThevenet]

Did you exchange with JSC on this? Any chance this could be useful feedback to them?

[AlexanderSinn]

Yes I talked to JSC support a lot about this. However, we agreed that the setting might not improve performance (and could worsen it) for most other codes.