Derive AcquisitionMetadata.n_channels from channel_types instead of storing it

[bruAristimunha]

Split out of #1161 so the mechanical part is reviewable on its own. Net −265 lines.

The problem

AcquisitionMetadata stored n_channels next to channel_types, which describes the same fact. The two could disagree, and for 34 of the 147 catalogued datasets they did.

The cause was not carelessness — the field was being used with two different meanings:

Dreyer2023A.METADATA.acquisition
#   n_channels   = 27                                  <- EEG electrodes only
#   channel_types = {"eeg": 27, "emg": 2, "eog": 3}     <- 32
#   sensors       = [... 32 entries ...]                <- 32

48 datasets declare more than one channel type, so the two readings diverge for a third of the catalogue.

The fix

Delete the field; derive it.

@property
def n_channels(self) -> int:
    return sum(self.channel_types.values())

It now consistently means every recorded channel — the reading that each dataset's own sensors list and test_n_channels_matches_raw_data (which counts every non-stim channel) already assumed.

The 34 disagreements are not fixed so much as made impossible, which is why this adds no validator and no invariant test. validate_metadata_against_dataset needs no new assertion.

What that removes

• 218 n_channels= arguments across 93 files
• the enrichment branch in metadata/__init__.py whose only job was to backfill the value
• two replace(...) defaults that also disagreed with their own channel_types

Behaviour change

For the 34 datasets that stored the EEG-only count, n_channels now reports the total. Dreyer2023A goes 27 → 32. One test asserted the old reading and now asserts the total; it is the only one in the suite that pinned a literal value for a multi-type dataset.

Documented under API changes, not Bugs: passing n_channels= to the constructor is no longer accepted.

Verification

• n_channels != sum(channel_types): 34 → 0, and now unrepresentable
• every AcquisitionMetadata(...) call site still passes sampling_rate and channel_types — checked by AST across the package, 0 broken
• 1740 passed (the single failure, test_lee2024_data_path_downloads_the_real_upstream_inventory, is pre-existing on develop)
• pre-commit clean

Closes #1163.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: d61a622314

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]

Remove n_channels from the remaining coverage fixture

Running pytest -q moabb/tests/test_bids_enrichment.py::TestReadmeAcquisitionFields now produces 17 failures because test_field_present still includes n_channels in the constructor kwargs at moabb/tests/test_bids_enrichment.py:2164. Since this property replaces the constructor field, update that parametrized coverage test to derive the count through channel_types (and remove or adapt its n_channels case) so the normal test suite can pass.

Useful? React with 👍 / 👎.