Generalizing the MHD example benchmark

[martymichal]

The MHD example benchmark (one in examples/MHD, and configuration and results in benchmarks/MHD) is currently difficult to run because most of its configuration is hard-coded. This includes absolute filepaths that are specific to the LUMI supercomputer and to the users that ran the benchmark there.

@nguu0123, @ToxPuro, would it be possible to update the code to be more configurable/general? I'd like to explore specific configurations for the benchmark to plan more concrete benchmarking scenarios. I'm not sure which one of you is responsible for this particular example, so I'm tagging you both.

[nguu0123]

I think @ToxPuro as a developer of Pencil-code knows more and can help you update the code to be more general. The configuration I created in the repo is just for documenting, not enough for a benchmark tool. We can discuss more f2f if you want, I'm sick right now but I think I will be better tomorrow so from Wednesday I will be in office