Normalise models BCI

[PierreGtch]

cf #457

[codecov]

Codecov Report

Merging #488 (1d53fca) into master (0f72195) will increase coverage by 0.29%.
Report is 1 commits behind head on master.
The diff coverage is 98.05%.

❗ Current head 1d53fca differs from pull request most recent head f4877ef. Consider uploading reports for the commit f4877ef to get more accurate results

@@            Coverage Diff             @@
##           master     #488      +/-   ##
==========================================
+ Coverage   83.50%   83.79%   +0.29%     
==========================================
  Files          60       61       +1     
  Lines        4426     4518      +92     
==========================================
+ Hits         3696     3786      +90     
- Misses        730      732       +2     

[cedricrommel]

LGTM. I checked the main EEGModuleMixin class, the deprecation function and 3-4 classes adaptations (not all of them).

[PierreGtch]

@robintibor @agramfort @bruAristimunha what do you think about replacing the model param ch_names by chs_info (a list of dict)? So that we can pass all the information available in epochs.info[‘chs’] to the model.