Add basic support for linting / format🐍

[KarelZe]

This PR adds basic support for linting/formatting based on ruff, as discussed in #23 .

A minimal rule set (see pyproject.toml) is appied to keep diff in PR small. More rules can be added from https://beta.ruff.rs/docs/rules/. Currently, I isort is used for sorting imports, pep8-naming N for naming conventions, NPY for Numpy best practices, and RUF for ruff-specific rules. You can also ignore rules, either in the pyproject.toml, as done for all lower-case variable names, which contradicts matrix naming conventions or on the code level e.g., # noqa: N802, as done in kernel_estimator.py for function_names due to the same reason.

We should verify that the changes from refactoring / update of numpy random number generation, didn't break anything. Will keep this as WIP until I've performed some tests 🤓

@iancovert Do you want additional rules / different configurations?

[iancovert]

All the ruff related changes look good to me!

For the random number generation, it looks like this was in just a couple places, right? I think the refactoring in iterated_estimator.py makes sense, but it could be better add a random_state argument to the constructor with a default argument. And the change in utils.py might be a problem, because the sample_subset_feature helper function would now return the same values every time. I think the correct solution is to make the random state an argument to sample_subset_feature, and adjust the two places where it's currently used (iterated_estimator.py and sign_estimator.py) to use their self.rng.

[KarelZe]

@iancovert Thanks for your feedback. I noticed the same behaviour for sample_subset_feature(...). This needs a fix. 🐛 Will also set random_state as a default argument.

Will update this PR and answer to #24 later this week. 👍

[KarelZe]

@iancovert After thinking about it again, I decided to revert my changes with regard to numpy.random and disable the rule NPY002, as it could affect the results estimated by the package. I feel like such a change should not be introduced in a linter/formatter PR and rather be adressed separately. Hope this is fine. 😊

I recommend to register the repo to https://pre-commit.ci/ (official pre-commit ci of pre-commit) to keep the hooks up-to-date.

From my point of view, PR is ready for review. 🫡

[iancovert]

Good call, that makes sense to me. I'll go ahead and merge the PR. Thanks again for adding this!

[KarelZe]

@iancovert Thanks for merging the PR. I might pick up work on a PR focusing on migrating to numpy 2.X (https://numpy.org/devdocs/numpy_2_0_migration_guide.html) and RNG, once I find the time.