Osl-dynamics

PyPi release: https://pypi.org/project/osl-dynamics/2.1.0/

**UPGRADED TO TENSORFLOW 2.19.** Older versions are not compatible.

Changes:
- Refactored models:
- Now output dicts.
- Moved `_model_structure` into the model class (`build_model`).
- Rewrote how the losses are logged during training (needed for Keras 3, where the implementation of metrics was changed).
- Rewrote optimizer for the HMM.
- Fixed bug in TFRecord datasets related to TensorFlow upgrade.
- Updated installation instructions.
- New conda environments.
- Removed pip dependencies. (`pip install osl-dynamics==2.1.0` won't install any dependencies.)

PyPi release: https://pypi.org/project/osl-dynamics/2.0.2/

**This is the last version compatible with Tensorflow v2.9-2.15.**

Changes:
- New fully TensorFlow HMM. This has significantly improved the efficiency of the HMM. **Not backwards compatible.**
- Removed dependency on `numba`.

PyPi release: https://pypi.org/project/osl-dynamics/2.0.1/

Changes:
- New function to calculate HMM-based features for downstream predictions.
- Improved models combining a state description with DyNeMo.
- Option to add extra channels to the Data object.
- Improved efficiency of HMM dual estimation.
- Unified initialisation methods.
- Fixed bugs.

PyPi release: https://pypi.org/project/osl-dynamics/2.0.0/

**This is the first version to be included in FSL.**

Changes:
- License changed from MIT to Apache-2.0 (needed to be changed for FSL).
- Option to stop HMM training early based on maximum fractional occupancy change.
- Removed WIP models and examples.

PyPi release: https://pypi.org/project/osl-dynamics/1.4.5/

Changes:
- New method for automated bad segment removal.
- Standalone GLM module.
- Support for loading mat73 files is deprecated.
- Various bug fixes.

PyPi release: https://pypi.org/project/osl-dynamics/1.4.4/

Changes:
- **Coherence calculation with `analysis.spectral.regression_spectra` modified: fixed issue with `nan`.**
- **Changed feature of validation split method of the Data object to split within sessions.**
- Option to save model checkpoint. This allows the user to resume training in a separate script.
- Added method to the Data object to align the sign of channels across sessions.
- Added new single-channel DyNeMo model (still a work in progress).