`KSComplement` can be unstable for constant float values

[fealho]

Environment Details

• SDMetrics version: 0.16.0

Error Description

The KSComplement may return a score of 0.0 when passed seemingly identical float values.

Steps to reproduce

from sdmetrics.single_column.statistical.kscomplement import KSComplement
import pandas as pd

vals1 = pd.Series([.2 - .1])
vals2 = pd.Series([.3 - .2])
print(KSComplement.compute(vals1, vals2)) # This prints 0.0 

Expectation

The KSComplement is expected to return 1.0 for such cases.

Additional context

This seems to be the default behavior of scipy's ks_2samp, which KSComplement uses to compute the KS test statistic.