Don't plot imputed data

We probably shouldn't plot imputed data as that creates artifacts. For example on 'breast-w':

from sklearn.datasets import fetch_openml
from dabl import plot
data = fetch_openml("breast-w", as_frame=True)
df = data.frame
plot(df, target_col='Class', type_hints={c: 'continuous' for c in df.columns if c != 'Class'})

However, determining which features are useful requires the imputed data. Maybe we should be passing around a more complex class that has more information?