ENH: Check CIFTI-2 data shape matches shape described by header

[MichielCottaar]

Checks whether the shape of the data matches the shape expected based on the CIFTI-2 header

• raises an error when trying to write out a mismatched CIFTI image to disk
• raises a warning when creating a mismatched CIFTI file
  I've added tests for both of these behaviors (and fixed many old tests that produced mismatched CIFTI images).

In the end I did not adopt the framework of _get_checks as this seemed to be only implemented for subclasses of WrapStruct, which are the header objects. However, this is a test that should be run on the Image class (as it requires knowing the shape of the dataobj).

[MichielCottaar]

I do think a ConsistencyWarning rather than a UserWarning would work well here, if you decide to implement one.

[codecov-io]

Codecov Report

Merging #774 into master will increase coverage by 0.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master     #774      +/-   ##
==========================================
+ Coverage   89.97%   89.99%   +0.01%     
==========================================
  Files          94       94              
  Lines       11992    12009      +17     
  Branches     2128     2134       +6     
==========================================
+ Hits        10790    10807      +17     
  Misses        859      859              
  Partials      343      343

Impacted Files	Coverage Δ
nibabel/batteryrunners.py	95.74% <ø> (ø)	⬆️
nibabel/cifti2/cifti2.py	96.55% <100%> (+0.1%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 0282bb9...c531421. Read the comment docs.

[effigies]

This looks good to me. Just a couple comments.

[effigies]

Also, I doubt I'm going to have time to think through and implement new warning classes before the 2.5 release. Perhaps we can revisit afterwards.

[MichielCottaar]

I'm happy to either have this pull request to remain open until you have time to decide on the new warning system. We could also just put it in with the default python warning and later replace that with a custom warning system. I have no strong preference either way.

[effigies]

Thanks. One quick typo...

[effigies]

LGTM. Thanks.