ENH: add LocalCrossPlot (Westerholt) plot and ability to scale Moran scatterplot by LOSH

[martinfleis]

xref #445

@ljwolf before I craft some simple tests and figure out a way of getting a reasonable legend, can you give this a pass?

Happy to iterate on the name, @rwesterh does not like the Westerholt plot :).

[codecov]

Codecov Report

❌ Patch coverage is 92.30769% with 8 lines in your changes missing coverage. Please review.
✅ Project coverage is 82.5%. Comparing base (cdb8076) to head (f6181cf).

Files with missing lines	Patch %	Lines
esda/inspection.py	89.7%	6 Missing ⚠️
esda/losh.py	66.7%	1 Missing ⚠️
esda/moran.py	95.8%	1 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##            main    #446     +/-   ##
=======================================
+ Coverage   82.2%   82.5%   +0.3%     
=======================================
  Files         27      28      +1     
  Lines       3969    4051     +82     
=======================================
+ Hits        3264    3342     +78     
- Misses       705     709      +4     

Files with missing lines	Coverage Δ
esda/__init__.py	100.0% <100.0%> (ø)
esda/losh.py	95.7% <66.7%> (+0.3%)	⬆️
esda/moran.py	85.2% <95.8%> (+0.3%)	⬆️
esda/inspection.py	89.7% <89.7%> (ø)

... and 1 file with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[ljwolf]

This is good! I think at minimum we need to allow a significance level set by the user. I also started down the path of implementing a multiple comparisons correction, but (reflecting on it) I think that should be done at the estimator level (through the new significance calculations prompted by #281)

[ljwolf]

plot_local_cross() might be a good name–--it's a cross-statistic plot (Gi, Ii, and Hi) and it's also a cross-shape.

[rwesterh]

Thank you @ljwolf and @martinfleis for implementing the plot! The main reason I don’t like the name ‘Westerholt plot’ is that the plot does not represent a statistical measure that I developed myself. The Moran scatterplot is not called ‘Moran scatterplot’ because Pat Moran proposed the plot, but because it visualises his Moran’s I statistic. I like @ljwolf’s suggestion to call the plot a 'local cross plot'. 'Cluster heterogeneity plot' would probably also be a good name, reflecting the main purpose of the plot, namely the visualisation of different dimensions of cluster heterogeneity.

[martinfleis]

Cool, I have renamed the class to LocalCrossPlot and also added the ability to scale normal Moran scatterplot by LOSH as suggested by @ljwolf in #445.