MRG: 0.18 deprecations

[larsoner]

This is all of the deprecations for 0.18 except for EDF stim channels. It's not 100% trivial there because of all of the tal ch business, and #5160. We might need to keep the stim_channel parameter there for just that purpose, and say it will only be used for BDF files that actually have a stimulus channel.

But let's do that separately. The rest of this is pretty straightforward. I put the auto covariance stuff in a context manager to ensure the the data are always un-scaled (this wasn't ensured before, e.g., if there was an error).

[codecov]

Codecov Report

Merging #5733 into master will increase coverage by <.01%.
The diff coverage is 95.72%.

@@            Coverage Diff             @@
##           master    #5733      +/-   ##
==========================================
+ Coverage    88.5%   88.51%   +<.01%     
==========================================
  Files         369      369              
  Lines       69367    69046     -321     
  Branches    11689    11638      -51     
==========================================
- Hits        61395    61116     -279     
+ Misses       5100     5068      -32     
+ Partials     2872     2862      -10

[massich]

Reviewing this PR is not easy. It would had been better to put everything in an issue and do smaller PRs. We'll know for next time. (I use #5737. Maybe you can add whatever you had in mind #5733 (comment)))

[massich]

I think I'll call it a day. @larsoner feel free to merge when green.

[massich]

ups! now codecov did complain

[larsoner]

Need to wait until all CIs return

[massich]

true, it needs to collect all the reports. my bad.

[larsoner]

@agramfort feel free to merge if you are happy

[agramfort]

awesome ! thanks @larsoner