Add support for ERP analysis: ignore_polarity = False

[vferat]

Closes #160

As discussed in #92, we could add support for taking into account polarities in the analysis.
Here is a list of changes that need to be implemented:

• Add self._ignore_polarity (bool) attribute to _BaseCluster . This attribute can only be changed using the ignore_polarity argument of the .fit() method ( to be added).
  By default during fitting, ignore_polarity should be set to True for Raw , Epochs , ChData and False for Evoked .
• .predict should have the same behaviour as .fit() : it should use self._ignore_polarity.
• ignore_polarity value used during prediction should be store in segmentations
• _BaseCluster.invert_polarity should not work when self._ignore_polarity is set to False.
• metrics should take into account cluster._ignore_polarity .
• I/O roundtrip!

Feel free to complete this list

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (ff3da2b) 90.00% compared to head (a3e5c12) 96.33%.

❗ Current head a3e5c12 differs from pull request most recent head 721bd45. Consider uploading reports for the commit 721bd45 to get more accurate results

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #93      +/-   ##
==========================================
+ Coverage   90.00%   96.33%   +6.32%     
==========================================
  Files          33       54      +21     
  Lines        2892     4121    +1229     
  Branches      678        0     -678     
==========================================
+ Hits         2603     3970    +1367     
+ Misses        172      151      -21     
+ Partials      117        0     -117     

Flag	Coverage Δ
unittests	?

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[mscheltienne]

Adding one more bullet point:

• I/O roundtrip!