NaN values need to be removed from grouping series in method filter_and_group_series

In some cases the DataFrames have column names like NaN returned from method filter_and_group_series (and also Unnamed 0 returned from method dataframe_value_counts) instead of column names resulting from the grouping series.

A workaround that removes those columns can be:

dataframe_time_per_center = \
    util.filter_and_group_series(series_coding_time,
                                 series_center.dropna())

Maybe the NaN column name comes from NaN values in the grouping series, hence it would by nice to have an option like drop_nan=True for method filter_and_group_series.