Do not use the deprecated matplotlib config option 'text.latex.preview'.

[csadorf]

Deprecated as of version 3.3.4, see:
https://matplotlib.org/3.3.4/api/api_changes.html#text-latex-preview-rcparam

Requires matplotlib~=3.3.

Fixes #5231.

[csadorf]

Alternative to #5232 .

[codecov]

Codecov Report

Merging #5233 (a3fae75) into develop (01a8846) will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff            @@
##           develop    #5233   +/-   ##
========================================
  Coverage    81.23%   81.23%           
========================================
  Files          533      533           
  Lines        37356    37356           
========================================
  Hits         30344    30344           
  Misses        7012     7012           

Flag	Coverage Δ
django	76.11% <ø> (ø)
sqlalchemy	75.19% <ø> (+0.03%)	⬆️

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

Impacted Files	Coverage Δ
aiida/orm/nodes/data/array/bands.py	77.46% <ø> (ø)

---

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 01a8846...a3fae75. Read the comment docs.

[csadorf]

@sphuber All tests pass with this patch, I am just not 100% sure whether there are any side-effects that are not captured by the tests. I will add @giovannipizzi as reviewer, since he originally authored this particular code segment.

[sphuber]

Thanks for the PR @csadorf . Would indeed be good to have @giovannipizzi give a look as I have never worked on nor used this code. Looking at the comment in our code and the deprecation message from matplotlib, it seems this option was necessary to provide correct alignment when using the TeX output format, but as of more recent matplotlib versions, that is done automatically. The ideal solution then would maybe to just have a conditional import. If an older version of matplotlib is installed, use the deprecated option, otherwise don't bother. Not sure how often this code is used though and so whether it is worth the effort. Pretty sure that the tests don't explicitly check this, so very possible that removing it will lead to incorrect images with older versions of matplotlib

[csadorf]

The ideal solution then would maybe to just have a conditional import. If an older version of matplotlib is installed, use the deprecated option, otherwise don't bother. Not sure how often this code is used though and so whether it is worth the effort. Pretty sure that the tests don't explicitly check this, so very possible that removing it will lead to incorrect images with older versions of matplotlib

I've added the matplotlib~=3.3 dependency with this patch, so I don't think there is a need need for a conditional import. However, I am just realizing that we should probably add ,>=3.4.4 to ensure that we are actually getting the right version.

[giovannipizzi]

I couldn't test it, but I'd say it's OK to remove it!
Thanks

[csadorf]

@chrisjsewell Alerting you to this fix which might warrant a hotfix release.