v0.1.1 - Split loading and deserializing / Project Documentation

Software Changes:

• Bump default dataset version: v0.0.3 → v0.0.4.
• Between v0.0.3 and v0.0.4 of srlearn/datasets, all zipfiles now have the version number appended
  (e.g. toy_cancer_v0.0.4.zip). Add logic to request the correct filename from GitHub.
• Add deserialize_zipfile function, split out code for pulling zipfile content from the load method.
• Add private _make_file_path function to handle where zipfiles are stored on a user's filesystem.
• Move RelationalDataset type into relational_datasets/types.py
• Fix hayesall/relational_datasets → srlearn/relational_datasets in setup.py
• Clarify typing.Optional in function signatures where default file paths are allowed.
• Add __version__ to the main __init__, so print(relational_datasets.__version__) is valid.

Documentation:

• Add mkdocs builds with each push to the main branch.
• Add requirements_dev.txt with requirements to build documentation.
• Add docs/build._docs.py to build a Downloads page and an overview of each dataset pulled from the srlearn/datasets repository.
• Add docs/notebooks/ directory for literate tutorials
  • Add 00_loading_and_fetching.ipynb
• Pages for functions and types:
  • types.RelationalDataset
  • request.deserialize_zipfile
  • request.fetch
  • request.latest_version
  • request.load