DOC: metadata-only paradigm compatibility recipe

[bruAristimunha]

Summary

• Add a metadata-only example using the existing public MotorImagery.is_valid API, without loading recordings, constructing evaluations/Results, enumerating datasets, or training.
• Explain explicit versus default n_classes semantics, one/two declared sessions versus actual data, unresolved remote metadata, and unknown evaluation compatibility.
• Add a focused regression that executes the example with file/network/data/evaluation/training entry points guarded and checks metadata, paradigm, and environment preservation.

Deliberately limited scope

This is a documentation recipe, not a new combined evaluation validator. Evaluation predicates are instance methods and construction initializes Results and filters the supplied dataset list. Rather than introduce unsupported dummy receivers or refactor those predicates, this reuses the existing public paradigm API and leaves evaluation compatibility null. No evaluation semantics, package API, dependencies, workflow, competition settings, or dataset loader changes.

Validation

• Recipe + existing paradigm tests: 103 passed (existing warnings only).
• Existing cross-session compatibility/reason and cross-subject compatibility tests: 3 passed.
• Dedicated no-I/O recipe regression: passed separately.
• All applicable pre-commit checks and git diff --check: passed.
• Independent source QA before push: APPROVE, no blocking defects; reviewer reran the focused test and reviewed all three files.
• Local Sphinx is unavailable; normal existing docs CI will build the gallery example. No laptop dataset downloads.

Normal required reviews/checks must pass before merge; no bypass requested.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 018e6beb2b

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Extend the reStructuredText title adornments

Sphinx Gallery parses this docstring as reStructuredText, but both adornments are 37 characters while the title is 38, so the documentation build emits a “title overline too short” parsing error and may render a system message instead of a proper gallery title. The section underlines on lines 28 and 50 are likewise one character shorter than their headings; extend all three affected adornments to match their heading lengths.

Useful? React with 👍 / 👎.