Test gen for serializers

[thorwhalen]

Proposed approach to test.

• Seeds: Make 1-3 dataframes covering different data types and structures we'd want to cover.
• Serializers: Make a list of several pandas serializers. These can be made as a partial (curry) of the combination of pandas serializers and their parameters.
• Going through the cartesian product of Seed and Serializers, serialize the seeds, within a try catch, collecting those that were valid.
• Save (as test files -- when applicable), including serializer (includes parametrization) in the filename somehow (use extension plus what ever is needed for parameters (example key=val, naming.
• Save seeds (in files or in code definition)
• Go through all serialized data and see if deserializers work.

See below, pairs of pandas serializers that have corresponding deserializers with the same name pattern.

for a in filter(lambda a: a.startswith('to_'), dir(pd.DataFrame)):
    aa = a.replace('to_', 'read_')
    if hasattr(pd, aa):
        print(f"{a:16} {aa}")

to_clipboard     read_clipboard
to_csv           read_csv
to_excel         read_excel
to_feather       read_feather
to_gbq           read_gbq
to_hdf           read_hdf
to_html          read_html
to_json          read_json
to_parquet       read_parquet
to_pickle        read_pickle
to_sql           read_sql
to_stata         read_stata