Clean Up pre/postprocessing

[davschneller]

This PR shall clean up the pre/postprocessing folders (if possible, move anything still needed into their own repositories), and remove some old scons files.

The following is done:

• site_scons has been removed
• The submodules scons-tools and cxxtest have been removed
• Documentation/UserManual was cleaned up, except for RD_sign_conventions.*.
• For preprocessing:
  • meshing: moved. See Port the preprocessing/meshing folder from the SeisSol repository to here Meshing#48 .
  • partitioning: removed (due to age). Only gambit.py was used by meshing/cube, so it has been copied to SeisSol/Meshing.
  • science: unsure what to do... (keep? Although extracting rconv would be good, I'd say. But where to put it?)
  • workflow: removed (due to age)
• For postprocessing:
  • performance: removed (due to age)
  • science: unsure what to do... (or where to put)
  • validation: unsure what to do... (or where to put)
  • visualization: moved. See Port from SeisSol/postprocessing/visualization Visualization#10 .
    Since there are some points for which I'm not sure what to do, this PR is still a draft.

[sebwolf-de]

We could leave rconv in the SeisSol repository for now, since it does not fit into neither meshing nor visualization.
I'd keep validation here or move it to precomputed-seissol. We need the scripts there for the tests.

[davschneller]

Personally, I think it would be good to have all compilable code but SeisSol itself out of the repo—at least long-term. I.e. we can also stop here for this PR and finish it.

More thoughts:

• In Port the preprocessing/meshing folder from the SeisSol repository to here Meshing#48 , Thomas suggested to turn the repositories into preprocessing (formerly Meshing and maybe Geodata) and postprocessing (formerly Visualization). ... Or should we maybe merge all of them into one? (we could also introduce a repository with submodules)
• I agree that rconv doesn't really fit into any of the existing repositories. Maybe, it would do best in a repository on its own, since the SeisSol/Examples repository uses it for e.g. WP2_LOH1. So why not make it like pumgen?
• That would leave only cube_c without its own repo (which is needed for the convergence tests).
• Also, it seems like gmsh2gambit is sort of obsolete at this point? The SeisSol/Examples repository goes directly from gmsh to pumgen. TODO: remove it from the cookbook then, I'd guess.

Or we can really leave the validation and science folders in here for now (since the latter have PRs for them). Maybe move rconv out, or not. Moving the validation to precomputed-seissol also seems good, although the scripts are not tied to the data there—so it's only maybe the right place for them.