Enable support for internally parallelized evaluation of the loss function using multi-rank workers

[mcw92]

This PR mainly implements multi-rank workers for internally parallelized evaluation of the loss function. In the future, this will be particularly useful for data-parallel training of individual neural networks during the Propulate optimization.

In addition the the intra-island communicator and the Propulate optimization communicator (which corresponded to MPI.COMM_WORLD before), the suggested implementation introduces another communicator at the level of each multi-rank worker, worker_sub_comm. This was realized by splitting MPI.COMM_WORLD reasonably in the islands.py module and also lead to changes with the other communicators before. Note that the worker sub communicator needs to be passed to the loss function as well to enable parallelized evaluation.
An example script multi_rank_worker_example.py is available in the tutorials/ folder.

In addition, the the following other things were fixed "on the fly":

• Clean up CMA-ES.
  • Write clean docstrings.
  • Remove unnecessary getter and setter functions.
  • Rename variables so their names speak for themselves.
  • Make serial CMA-ES reproducible by passing the Propulate RNG correctly.
• Improve README:
  • Add comment about new Propulate mailing list and GitHub discussions.
  • Add minimum working example to Quickstart section (see also tutorials/minimum_working_example).
• Fix docstrings and type hints whenever I encountered things to fix.
• Basic update of ReadTheDocs pages:
  • Add minimum working example to Quickstart section.
  • Update simple Propulator example with logging functionality.
  • Minor changes for the islands and HPO tutorial.
  • Add a stub for the multi-rank worker tutorial.

Things to do and fix in future PRs:

• Write integration test for multi-rank worker functionality. This will be done once the already available tests are transformed to MPI-enabled tests.
• Add tutorials for PSO, CMA-ES, and multi-rank workers to ReadTheDocs.

Closes #24, #102, #107