* ENTRY_SEED: A dict with entry data that can be assumed when it's

      coming from this source.

    * _read_header(cls, input_file): Some CSVs include "headers" with smaller

      rows before they get to the "real" data.  This classmethod is expected to

      read those rows and return two values: a dict of entry data read from

      the headers, and a list of column names for the real data.  The method

      is expected to leave input_data at the position where the real data

      starts, so callers can run ``csv.DictReader(input_file, column_names)``

      after.

      The default implementation reads rows until it finds one long enough to

      include all of the columns required by NEEDED_FIELDS and COPIED_FIELDS,

      then returns ({}, that_row).

    * _read_header_row(cls, row): A classmethod that returns either a dict,

      or None.  The default implementation of _read_header calls this method

      on each row.  If it returns a dict, those keys and values will be

      included in the entry data returned by _read_header.  If it returns

      None, _read_header expects this is the row with column names for the

      real data, and uses it in its return value.