tests for dots broken for python 3.8 3.9 with "last lambda bin is not empty"

[sergpolly]

this is all due to pandas-dev/pandas#48645

and is already fixed in pandas 1.6.dev

that's where is breaks with pandas 1.5.0:

cooltools/cooltools/api/dotfinder.py

Line 897 in a5341aa

scored_df.groupby([obs_raw_name, lbins], dropna=False, observed=False)[

if curious check minimal reproducible example:

import numpy as np
import pandas as pd

data = pd.DataFrame({"a":[0,0,0,1,1,2]})
bins  = np.r_[-np.inf,[1,2,3,4],np.inf]
categories = pd.cut(data["a"],bins)
# outputs below should be identical - but they are not with pandas 1.5.0
display(data.groupby(categories,dropna=True).count())
display(data.groupby(categories,dropna=False).count())

[Phlya]

I think the freshly released pandas 1.5.1 has this bugfix included

[gfudenberg]

with

a = data.groupby(categories,dropna=True).count()
b = data.groupby(categories,dropna=False).count()
a.equals(b)

in 1.5.1 !