Sorting by a categorical column doesn't always work

[rjzamora]

While debugging a dask-expr + cudf issue, I discovered that sort_values does not always work when by is a categorical column (regardless of whether query-planning is enabled).

Reproducer:

import numpy as np
import pandas as pd
import dask.dataframe as dd

df = pd.DataFrame()
df["a"] = np.arange(10)[::-1]
df["b"] = df["a"].astype("category")
ddf = dd.from_pandas(df, npartitions=2)
got = ddf.sort_values(by="b")
expect = df.sort_values(by="b")
dd.assert_eq(got, expect, check_index=False)

As far as I can tell, problem is that in places like this or this. It seems that we cannot always call min on a categorical series.