Roerich

:cyclone: Updates:
- `EnergyDistanceCalculator ` method is added for change point detection based on energy distance calculation.
- `ChangePointDetectionClassifierCV` added as a version of `ChangePointDetectionClassifier` method with K-Fold cross-validation.
- `MatrixImportance` added to calculate a matrix of importance for each pair of input features in change point detection task.
- Now we use automatic tests for all methods in `roerich`.
- Methods are divided into different modules for future versions. Compatibility with v0.4.0 is maintained.

:page_with_curl: Documentation:
- Added descriptions of all new methods in `roerich`.

:crystal_ball: Documentation:
- Added descriptions of all methods for change point detection in `roerich`
- Descriptions for density ratio estimation methods prepared
- Basic usage examples updated and published
- Several examples on real data demonstrated

:sparkler: Updates:
- `predict_proba_ratio(X)` added to density ration estimation methods
- `precision_recall_curve` function fixed
- `display` function updated

Changes

- Offline change point detection algorithms were implemented. The algorithms are described in https://doi.org/10.1016/j.jocs.2021.101385.
- Added quality metrics for change point detection results.
- Basic usage example was updated.
- `examples/demo.ipynb` was updated.
- The library description in README was changed.
- Dependencies were added to `setup.py`.
- Poetry installation is available.
- Related libraries for change point detection are listed in the README.

Notable Changes:
- removed RuLSIF
- added examples
- OnlineNNClassifier and OnlineNNRuLSIF are in roerich.algorithms now

This is the first release :)