df.groupby(df.group) produces duplicate column for some aggregation functons

It should be possible to use `df.groupby(df.group)` or `df.groupby('group')` and get the same result. Unfortunately for some aggregation functions (max, min, all, any), the former produces an output with an extraneous 'group' column. Note this doesn't happen for some functions, like size.

In groupby, we should check if the the series is one of this dataframe's columns when setting the index: https://github.com/apache/beam/blob/cdb882d9ae554556156bff4843f18567b214df13/sdks/python/apache_beam/dataframe/frames.py#L156

Imported from Jira [BEAM-11305](https://issues.apache.org/jira/browse/BEAM-11305). Original Jira may contain additional context.
Reported by: bhulette.