Fix warnings from scikit-learn 1.8+ and introduce FixedPipeline

[bruAristimunha]

Update the preprocessing logic to address warnings from the latest scikit-learn version. Introduce a FixedPipeline class to handle pipelines that do not require fitting, ensuring compatibility with future versions of scikit-learn. Adjust existing code to utilize FixedPipeline where appropriate.

[Copilot]

Pull request overview

This pull request addresses warnings from scikit-learn 1.8+ by introducing a FixedPipeline class that is always considered fitted, eliminating FutureWarning messages for unfitted pipelines. The changes ensure that preprocessing pipelines that only apply fixed transformations (filtering, epoching, etc.) are properly marked as fitted.

Key Changes

• Introduced FixedPipeline class and make_fixed_pipeline function that create pipelines marked as fitted via the __sklearn_is_fitted__ method
• Updated all FixedTransformer subclasses to properly call super().__init__() and return self from fit() methods
• Replaced Pipeline with FixedPipeline and make_pipeline with make_fixed_pipeline throughout the preprocessing and paradigm code

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.

File	Description
moabb/datasets/preprocessing.py	Introduces FixedPipeline class and make_fixed_pipeline function; adds sklearn_is_fitted methods to ForkPipelines and FixedTransformer; ensures all FixedTransformer subclasses call super().init()
moabb/paradigms/base.py	Replaces Pipeline with FixedPipeline and make_pipeline with make_fixed_pipeline; updates imports to include the new FixedPipeline utilities
moabb/datasets/base.py	Replaces Pipeline with FixedPipeline in dataset processing pipeline creation and cache handling
docs/source/whats_new.rst	Documents the fix for scikit-learn warnings in the changelog

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

The parameter documentation is incorrect. The function uses _name_estimators(steps) which means it expects bare estimators, not tuples. The docstring should say:

*steps : list of estimators
    List of estimators that are chained in the pipeline. These are automatically named.

This matches sklearn's make_pipeline behavior, which takes estimators directly rather than named tuples (unlike the Pipeline constructor).

Suggested change

	List of (name, transform) tuples that are chained in the pipeline.
	List of estimators that are chained in the pipeline. These are automatically named.

[toncho11]

Thanks!!! That message was terrible! It was flooding the console to the point that I could not see any other message and I had to revert to an older version.