loading ERM fails when preparing for maxfiltering if ERM contains EEG

[drammock]

pipeline calls to maxwell_filter_prepare_emptyroom seem like they'll always fail when the ERM contains EEG channels, because the pipeline always removes everything except MEG channels from raw_ref:

mne-bids-pipeline/mne_bids_pipeline/_import_data.py

Lines 469 to 480 in 9d2548b

	raw_ref.pick_types(meg=True, exclude=[])
	if prepare_maxwell_filter:
	# We need to include any automatically found bad channels, if relevant.
	# TODO: This 'union' operation should affect the raw runs, too,
	# otherwise rank mismatches will still occur (eventually for some
	# configs). But at least using the union here should reduce them.
	raw_er = mne.preprocessing.maxwell_filter_prepare_emptyroom(
	raw_er=raw_er,
	raw=raw_ref,
	bads="union",
	)

under the hood, maxwell_filter_prepare_emptyroom does this:

https://github.com/mne-tools/mne-python/blob/1d2635f84a55785c3531cfe4027eda3820a7fb31/mne/preprocessing/maxwell.py#L175-L176

So the result looks like:

E   ValueError: DigMontage is only a subset of info. There are 60 channel positions not present in the DigMontage. The channels missing from the montage are:
E   
E   ['EEG001', 'EEG002', 'EEG003', 'EEG004', 'EEG005', 'EEG006', 'EEG007', 'EEG008', 'EEG009', 'EEG010', 'EEG011', 'EEG012', 'EEG013', 'EEG014', 'EEG015', 'EEG016', 'EEG017', 'EEG018', 'EEG019', 'EEG020', 'EEG021', 'EEG022', 'EEG023', 'EEG024', 'EEG025', 'EEG026', 'EEG027', 'EEG028', 'EEG029', 'EEG030', 'EEG031', 'EEG032', 'EEG033', 'EEG034', 'EEG035', 'EEG036', 'EEG037', 'EEG038', 'EEG039', 'EEG040', 'EEG041', 'EEG042', 'EEG043', 'EEG044', 'EEG045', 'EEG046', 'EEG047', 'EEG048', 'EEG049', 'EEG050', 'EEG051', 'EEG052', 'EEG053', 'EEG054', 'EEG055', 'EEG056', 'EEG057', 'EEG058', 'EEG059', 'EEG060'].
E   
E   Consider using inst.rename_channels to match the montage nomenclature, or inst.set_channel_types if these are not EEG channels, or use the on_missing parameter if the channel positions are allowed to be unknown in your analyses.

I'm hoping / assuming that just removing EEG channels from ERM also is the right way to go here (?) If so I can open a PR.

[larsoner]

Yeah in principle EEG channels shouldn't be in empty-room data I think