Add support for discrete variables in rank plots

[aloctavodia]

Use a similar trick to the one used in plot_bpv for discrete variables.

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

Good sample

OLD

NEW

Bad sample (excess of 2s in one chain)

OLD

NEW

Bad sample (two different distributions)

OLD

NEW

As you can see with the old version discrete samples always look bad (and more or less the same). With the new version not only they look good when they should, they actually reflect the nature of problem with the sampling went wrong.

[codecov]

Codecov Report

Merging #1433 into master will increase coverage by 0.00%.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##           master    #1433   +/-   ##
=======================================
  Coverage   91.91%   91.92%           
=======================================
  Files         105      105           
  Lines       11207    11216    +9     
=======================================
+ Hits        10301    10310    +9     
  Misses        906      906           

Impacted Files	Coverage Δ
arviz/plots/backends/bokeh/rankplot.py	94.59% <100.00%> (-0.08%)	⬇️
arviz/plots/backends/matplotlib/rankplot.py	96.92% <100.00%> (-0.05%)	⬇️
arviz/plots/plot_utils.py	91.49% <100.00%> (+0.39%)	⬆️

---

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 9c21881...2fa6b69. Read the comment docs.