Improve documentation on "impute_function"

[AsiehH]

I receive an error when passing a dataframe with nan in some columns in the following code where df_rolled is the output of roll_time_series:

from tsfresh.feature_extraction import extract_features, EfficientFCParameters
from tsfresh.utilities.dataframe_functions import impute

extraction_setting = EfficientFCParameters()
X = extract_features(df_rolled, column_id="id", column_sort="month", default_fc_parameters=extraction_setting, impute_function=impute, show_warnings=False)

ValueError: Column must not contain NaN values: ratio_last_sum_PPDOS

There is an old issue #18 but it appears that tsfresh.feature_extraction. FeatureExtractionSettings doesn't exist anymore.

[AsiehH]

Ok, so I figured out that the impute_function parameter in extract_features does not impute missing values in the input but the output!
It would be great to clarify this in the documentation here

Currently the definition is as follows

impute_function

(None or callable) – None, if no imputing should happen or the function to call for imputing>

[nils-braun]

Thanks for getting back to the issue and looking for the solution by yourself. If you have a a good suggestion, we would love to see a PR :-)
Otherwise we can also try to fix it