Esnpy

- :sparkles: `SklearnTrainer`

- :recycle: major refactoring
- `ReservoirConfig` is now `ReservoirBuilder` with a `build()` method
- setup `__all__` everywhere
- :memo: sphinx documentation alpha
- :rocket: documentation publishing on release

- :lipstick: PyPI project's url in pyproject.toml

- :recycle: Trainer refactoring
- add `use_input` abstract method
- remove `_Identity` class
- remove dirty None trick in DeepESN configs
- :rocket: improved github workflow
- :memo: improved readme
- better quickstart
- move roadmap section to github projects

- :recycle: Awesome refactoring
- :sparkles: Features
- model: ESN, DeepESN
- init: Uniform or Normal distribution, sparse or dense matrix
- tune: SpectralRadiusSetter
- train: RidgeTrainer
- :memo: Examples
- Mackey-Glass prediction
- 2D trajectories classification

Legacy implementation