Oodeel

**Features**
- New OOD detectors with tests and tutorials:
- **RMDS**: Added a new method for OOD detection.
- **GEN (Generalized Entropy)**: Implemented a new OOD detector with tests and documentation.
- **SHE (Simple Hopfield Energy)**: Added SHE baseline with tests and tutorial notebooks.
- **SCALE and ASH**:
- **DataHandler Enhancements**:
- Added `return_tuple` option to `data_handler.prepare` for flexible return types.
- Added backend initialization to `TFDataHandler` and `TorchDataHandler`.
- Introduced `data_handler_loader` to load the appropriate `DataHandler` based on the backend.
- **DkNN**: Added support for `faiss-gpu`.
- **Feature Extractors**: Added a verbose option for better logging.

**Refactors**
- Renamed variables and methods for clarity:
- `feature` and `keys` renamed to `column_name` and `columns` in `data_handler`.
- `data_handler_loader` renamed to `load_data_handler`.
- `_clean_hooks` replaced with `_clean_ood_hooks`.
- Deprecated and moved old `data_handler` and `OODDataset` files to a separate directory.
- Removed unused attributes and methods (e.g., `_ood_handles`, feature assignment methods).

**Fixes**
- Fixed various issues:
- Fixed compatibility issues with NumPy and TensorFlow.
- Fixed bugs in methods like Mahalanobis
- Fixed docstrings and clarified default argument values.
- Fixed environment issues for tests and coverage.

**Tests**
- Added or updated tests:
- Tests for deprecated `DataHandler` and `OODDataset`.
- Adapted tests to new terminology and refactored methods.
- Updated tests for React and other modules.

**Documentation**
- Improved documentation:
- Added tutorial notebooks for new methods (e.g., GEN, SHE, SCALE, ASH).
- Unified and executed demo notebooks.
- Updated README with new features and badges.
- Improved docstrings for various methods and classes.
- Added versioning for documentation using `mike`.

**Chores**
- Code formatting and environment updates:
- Added `ruff` formatter and adjusted line width.
- Fixed Flake8 issues and typos.
- Updated dependencies and environment configurations (e.g., added `tqdm`, `tox-uv`, fixed NumPy compatibility).

**Build**
- Updated build and environment:
- Added attributes to relevant baselines.
- Fixed development and test environments.
- Added `requests` as a requirement for GitHub Actions.

This summary captures the key changes across features, refactors, fixes, tests, documentation, chores, and build updates. Let me know if you need more details on any specific part!

* New baseline: Gram! [Detecting Out-of-Distribution Examples with Gram Matrices](https://proceedings.mlr.press/v119/sastry20a.html), ICML 2020. Check out the tutorial to know more :).
* FeatureExtractor.predict() and FeatureExtractor.predict_tensor() now takes a new argument as input: postproc_fns. This argument is a list of Callable functions that applies to the batch features obtained after each (batch-wise) forward. It allows us to perform postprocessing on internal features on the fly rather than after a forward pass on the whole dataset that would require saving the features of the whole dataset in GPU memory, likely leading to an OOM (especially for feature maps).

First Release on Pypi!
Check out the doc and Readme to get to know oodeel!

This release is stable, but still beta since although usable and self-contained, it does not encompass all the end-goal features that we plan to implement.