Fix bug in pair plot axis handling

[wm1995]

Description

This is an attempt at fixing #1981, in which the subplot axes do not have the same range by default but also lack unique labels. In doing so, I hope to correct some of the counterintuitive behaviour around passing "sharex" or "sharey" as backend parameters. I've written a fix, and am in the process of adding unit tests for said fix and unit tests for the fix.

Update (23/02/22):
I generated an example plot of the marginals, based on the plotpair example gallery code, to illustrate that the marginal axes now fall in the correct place.

Cf. the old behaviour, where the marginal plot axis ticks are not aligned with the plots beneath them, but the labels are also missing.

Checklist

• Follows official PR format
• Includes a sample plot to visually illustrate the changes (only for plot-related functions)
• Includes new or updated tests to cover the new feature
• Code style correct (follows pylint and black guidelines)
• Changes are listed in changelog

(I removed the "new features" documentation from the list because this isn't a new feature, just a bug fix to correct unexpected behaviour)

[codecov]

Codecov Report

Merging #1985 (b6fcefa) into main (9d74000) will increase coverage by 0.00%.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##             main    #1985   +/-   ##
=======================================
  Coverage   90.83%   90.84%           
=======================================
  Files         114      114           
  Lines       12311    12321   +10     
=======================================
+ Hits        11183    11193   +10     
  Misses       1128     1128           

Impacted Files	Coverage Δ
arviz/plots/backends/matplotlib/pairplot.py	92.34% <100.00%> (+0.41%)	⬆️
arviz/plots/densityplot.py	91.07% <0.00%> (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 9d74000...b6fcefa. Read the comment docs.

[wm1995]

Right, I think this PR is ready for review now. The test I put together isn't strictly necessary, I think, but it is useful to make sure that all the axes you'd expect to be shared are indeed shared.

[wm1995]

Another example of the fix introduced by this PR is in the plot_pair_point_estimate example - cf. the original and the updated pages in the docs - the point estimates now actually line up in the marginal distribution and the joint distribution plots.

[OriolAbril]

Thanks for the improvement

[OriolAbril]

I left two minor comments, thank you so much for your contribution and patience while we discussed the approach to implement.

[wm1995]

I've made the requested changes, will just make sure that the CI runs alright and then if all is OK I'll leave it with you. Thanks for your help in putting this together and for being so responsive!