Add support for MNE-ICALabel

[hoechenberger]

Before merging …

• [x] Changelog has been updated (docs/source/changes.md)

[hoechenberger]

@larsoner I'm not sure why this one test fails, could you help me out here please? :) https://github.com/mne-tools/mne-bids-pipeline/actions/runs/6798018175/job/18481346112?pr=812

[hoechenberger]

• TSV of detected ICs
• ICA blink removal – there's obviously one huge EOG artifact that was not detected, possibly because auto reject-local kicked out too many epochs before fitting ICA?
• ICA cleaning of experimental data
• Final report – expected effect is found (N170)

[larsoner]

It only looks for items that are ast.AnnAssign so it's because you did not type annotate. I added a test that will check this more explicitly and fixed the error

[hoechenberger]

Thanks @larsoner!!

[larsoner]

Also @hoechenberger we are almost through all of our 250,000 paid CircleCI compute credits for Nov mostly from MNE-BIDS-Pipeline usage (141,856 MBP vs 46,692 for MNE), so if there is any way you could push less often and do more local testing it would be greatly appreciated :pray:

[hoechenberger]

Argh ok will try

[behinger]

hey! I tried to merge the newest changes in mne_bids_pipeline into this branch to make use if icalabel - but it is more tricky than anticipated because quite some code got moved around and I am not familiar with the code-base.

Any chance you have some newer local version already merged? If not, I will dig deeper into the code-base

Cheerio!

[hoechenberger]

Hey @behinger, we did some serious refactoring of the ICA-related project organization. I believe it would be easier to start a new PR based off of the latest main branch than trying to fix this one…

[behinger]

Okay, thanks for the response! We're (with @jschepers) looking into it :-)