Additional methods and fixes to DataFrameMixin

[adamamer20]

This PR adds some useful methods to DataFrameMixin and fixes the logic of some methods in pandas and polars. Added methods:

• remove: remove values from DF based on a mask
• add: equivalent of df.add from pandas
• all: equivalent of df.all from pandas
• column_names
• div: equivalent of df.div from pandas
• drop_columns
• drop_duplicates
• filter: remove values with a boolean dataframe
• groupby_cum_count: equivalent of df.group_by.cum_count from pandas
• join: equivalent of df.join from pandas
• mul: equivalent of df.mul from pandas
• norm: obtain a series with the L2 norm per every row in the DF
• reset_index: equivalent of df.reset_index in pandas
• set_index: equivalent of df.set_index in pandas
• with_columns: equivalent of df.with_columns in polars
• contains: equivalent to srs.contains in pandas
• range: get the series of a specified range (start, end, step)
• srs_to_df: convert series to DF

[rht]

Tangent: it seems to me it could be its useful own package for a general DF with either pandas or Polars backend. Useful for people who can't decide which one to use, like us, haha. Otherwise, to add a selling point, we could implement R's dplyr-style grammar of data manipulation. Though there is already https://github.com/machow/siuba.

[adamamer20]

Agree! A separate package with a unique interface for any DF library could be worthwile, especially since there are methods which are almost the same with only a name change. However right now I wrote up just the methods I needed so far. We can think about it when the library is a bit more mature.