MRG, ENH: Remove 15-char limit for FIF

[larsoner]

• Get things working with the block-based method
• Get other projects on board (taken care of in ENH: Remove 15-char limit for FIF #8348)
• Get ENH: Add channel block information fiff-constants#33 merged
• Update PR with newer fiff-constants commit, check 'bads' handling/replacement
• Add test where a file is written with long names, then monkeypatch to make the tag empty on read, ensure it can still be written and read
• Add to MNE-MATLAB

Closes #8348
Closes #8312

[larsoner]

@agramfort this is ready for review from my end. Once we're happy with this, I can make the corresponding changes in MNE-MATLAB.

Should we try to get this into 0.22, or make it an early 0.23 PR? I'm leaning toward the latter so we at least get a few months in of dev use

[agramfort]

let's merge this after the release

[larsoner]

Agreed. @bloyl would you be interested in doing the companion PR to MNE-MATLAB? My MATLAB brain has atrophied over the last many years of using Python...

[larsoner]

Companion PR is in mne-tools/mne-matlab#21, so these two are ready for review/merge from my end.

[agramfort]

works great !

just a few suggestions to make the code more self explanatory.

thx heaps @larsoner !

[agramfort]

it took me some time to understand that's why I suggested to rename to ch_names_mapping ? also is there a way to hide this for any public function?

[larsoner]

it took me some time to understand that's why I suggested to rename to ch_names_mapping

I'm not sure what you're referring to here -- I made this change...?

Pushed a change to preserve the API of read_ctf_comp. Failures are just pip pre

[agramfort]

thanks @larsoner !