Demonstrate a hack-y way to support Pint<->SQL interop

pd.to_sql can deal with Pint Pandas dataframes (provided all columns are some kind of pint[] type.  But reading back the dataframe requires quite a bit of unpacking.

Addresses (somewhat) https://github.com/os-climate/data-platform-demo/issues/50

Author: MichaelTiemannOSC