Update to use pyproject.toml for packaging metadata

[davidwilby]

📝 Description

At the moment, dependencies are specified in both setup.cfg and in requirements files under requirements/ which is a little odd and opens up potential errors such as in #131 where I bumped the version of neuralprocesses in the requirements file but not in the setup.cfg - the tests pass because they use the requirements file, but the built package would not have the same dependency version specification.

In this PR, I suggest moving to using pyproject.toml for specifying package metadata. This does solve the above problem, though that could also be solved using setup.cfg, but the python community seems to be converging on pyproject.toml. If we need to keep setup.py for any reason, that is also possible.

This PR also updates the workflows and docs accordingly.

I've tested that builds succeed and all seems to work as expected.

Feel free to close this @tom-andersson if it isn't to your taste.

✅ Checklist before requesting a review

(See the contributing guide for more details on these steps.)

• I have installed developer dependencies with pip install -r requirements/requirements.dev.txt and running pre-commit install (or alternatively, manually running ruff format before commiting)

If changing or adding source code:

• tests are included and are passing (run pytest).
• documentation is included or updated as relevant, including docstrings.

If changing or adding documentation:

• docs build successfully (jupyter-book build docs --all) and the changes look good from a manual inspection of the HTML in docs/_build/html/.

[tom-andersson]

This is great!! LGTM after my one comment is resolved.

I think there was a reason I went with setup.cfg rather than pyproject.toml when I set the deepsensor repo up 1.5 years ago, but I can't remember. I did read that pyproject.toml was starting to gain precedence, but it's probably even more encouraged now. If everything is working I'm happy because this is cleaner.

P.S. the pip install deepsensor[<deep learning library>] is great.

[tom-andersson]

Thanks @davidwilby!