Also I would recommend git grep _merge_grad_data'ing to see where else you might need to change things.

Codecov Report

Merging #7414 into master will increase coverage by 0.00%.
The diff coverage is 100.00%.

@@           Coverage Diff            @@
##           master    #7414    +/-   ##
========================================
  Coverage   90.03%   90.04%            
========================================
  Files         454      454            
  Lines       81400    82297   +897     
  Branches    13001    13020    +19     
========================================
+ Hits        73288    74103   +815     
- Misses       5267     5369   +102     
+ Partials     2845     2825    -20     

Also I would recommend git grep _merge_grad_data'ing to see where else you might need to change things.

So it appears _merge_grad_data is also used in get_peak, plot_projs_topomap, _plot_ica_topomap, plot_ica_components, plot_tfr_topomap, plot_epochs_psd_topomap and like a dozen helper functions.

So I think I will tackle each topic in turn to keep things bite sized. I will look at psd_topomap, ica, tfr, then projs in subsequent PRs. But first I will take and upload a short test measurement with overlapping channels so that I can make useful tests. @larsoner is this plan ok?

So I think I will tackle each topic in turn to keep things bite sized.

Sounds reasonable to me. Another approach might be to make _merge_grad_data a more general _merge_channel_data that handles both channel types, but based on its arguments I think this approach will not be straightforward (even if it ends up the way we go after you change a few of these code paths over). +1 for working in bite sized increments and we can see if patterns are repeated and can refactored.

#7057

@agramfort and @larsoner could you please review. This is working nicely for me.