implement cuda.jl version of: `rmg_minimal_rms_example.jl`

[JBlaschke]

This will also be helpful for understanding CUDA: https://www.olcf.ornl.gov/cuda-training-series

The code has a lot of reduction patterns, so this might be helpful:

• slides: https://www.olcf.ornl.gov/wp-content/uploads/2019/12/05_Atomics_Reductions_Warp_Shuffle.pdf
• video recording: https://vimeo.com/419029739

[JBlaschke]

For NVTX annotations (i.e. to see what's happening on on the CPU from nsight systems): https://juliagpu.github.io/NVTX.jl/dev/ which is part of CUDA.jl https://cuda.juliagpu.org/stable/development/profiling/#Source-code-annotations