Add Parquet I/O

[JSchlensok]

PR Checklist

• This comment contains a description of changes (with reason)
• Referenced issue is linked
• If you've fixed a bug or added code that should be tested, add tests!
• Documentation in docs is updated

Description of changes

Added module for reading/writing Pandas dataframes from/to Parquet files and partitioned datasets.

Technical details

Pretty straightforward. Deviated from the existing functionality for HDF5 slightly to enable support for multiple datasets indexed by name similar to HDF5's key-value structure through Parquet partitioned datasets.

Additional context

[JSchlensok]

Will investigate failing checks, some of them are definitely on me for forgetting to run pre-commit hooks 🌚

[JSchlensok]

@picciama All tests for Python 3.10 except safety (which reports vulnerabilities in tqdm and jinja2, which are not affected by this PR) run on my device - should I worry about the failing 3.8 test cases?

[picciama]

Oktoberfest will require at least python 3.9 so you can drop support for 3.8. Please change the requirements and classifiers in pyproject.toml accordingly, and update the matrix in the workflow file in .github/workflows/run_tests.yml so that the tests run using python 3.9. This will hopefully fix your issues. Then, please check mypy, there is also some issue there because you don't have a capital L for a list type somewhere I think. Safety issues may disappear when using a newer version of the packages that may not be available in python 3.8, hence it works on your machine with python 3.10 but not with 3.8 here. If safety is still failing, consider updating the dependencies that cause the issue.

[JSchlensok]

safety still reports https://data.safetycli.com/v/70612/97c/, for which there is no fix available currently and that is reported to be believed to be invalid by its maintainer. What is the best practice to go about that?

[picciama]

a few things require a second thought. Please fix. I fixed all remaining failing tests already.