Extend test suite to run more tests also in pre-commit

[terhorstd]

The test framework uses some outdated versions that start to become difficult with newer python versions. The main change in this PR is the update of some pyproject.toml sections and the .pre-commit-config.yaml. New checks are added to improve developer-local checks as well as the CI side.

In more detail:

• The mypy.ini was translated to pyproject.toml to have config in one place
• More pylint errors are silenced for now to have them being fixed in a separate PR.
• nest-server code was slightly refactored to meet absolute minimum quality requirements.
• some files with implicit relative import require explicit from . import connect_test_base to be checked. (which is subsequently resorted by isort)
• Some test implementation type-checking suffers from untyped pynest API, thus requiring explicit annotations like n_events: int = spike_recorder.get("n_events")
• mpi-tests can only be checked if directory-names are valid python module names, therefore the rename of mpi/2/*.py → mpi/nproc2/*.py
• Some updates caused a re-formatting of cpp file indention levels. (change "files changed" settings to hide white-space changes to de-clutter the review)
• minor changes to shell-scripts as suggested by shellcheck

[gtrensch]

LGTM! 👍 Approved.

[heplesser]

@terhorstd Thanks for this PR. Unfortunately, it comes with several changes that I believe hurt code readability, that I do not understand and that break the testsuite.

Concerning code formatting for C++, I think the changes to for-loop header indentation reduces readability. Even worse is the the line break after :: in many method declarations. Here, we need to stick to the old style.

I also do not understand why all tests suddenly are converted to importable Python packages. As far as I understand the Pytest documentation, when keeping tests outside the installable package as we do, tests should not be packages (https://docs.pytest.org/en/7.1.x/explanation/goodpractices.html#tests-outside-application-code).

All tests currently fail because connect_test_base.py is not found when tests try to import it.

[heplesser]

Here, rank is a global variable, is it not? Can we avoid that, maybe by restructuring the code?

[heplesser]

The else block is very large. Could one restructure the code to avoid such a long else block? Maybe return at the end of the if block?

[heplesser]

This reformatting looks very strange to me, significantly reducing code readability. Can we keep the old form and get the formatter to understand that this should be one line? Breaking on :: seems really weird to me.

[heplesser]

I do not like this change. I think the loop is more readable if all parts are aligned under each other as in the old version here. Can we not keep the old rules?