Version 2.0

[pietronvll]

This is a major release of kooplearn, which fundamentally changed its api to make it sklearn-compatible. The detailed changelog will be released upon merging. For now, the release is usable, but several tasks need to be completed before merging.

• Some tests of test_sklearn_compatibility.py.
• Remove .risk and .svals method in classical estimators and replace them with the sklearn-standar .score method.
• Add tests for kooplearn.torch and kooplearn.jax.
• Add compute_mueller_brown_potential_eig method in kooplearn.datasets.
• Go through the example notebooks and add commentary where missing (e.g. Ordered MNIST).
• Add the logistic map example.
• Go through every written docs and check for errors/typos (NO LLMs here).
• Convert the primer into a notebook, similarly to POT's quickstart guide
• Add github CI to run tests before merging and periodically.
• Add github CI to release on pip when merging onto main.
• Refresh read the docs documentation with the new version.
• JOSS paper.