Add parquet data source for integration testing

[jtcohen6]

Describe the feature

We've got people.csv and people.json, why not people.parquet?

Describe alternatives you've considered

Other formats we could consider adding, recognizing that they may not be supported on all databases:

• avro
• delta

Additional context

All databases should have support for parquet, since it's the canonical open source column-store format

Who will this benefit?

While this package can be used to stage CSV and JSON sources, the optimal production use case is columnar file formats + analytic databases/lakes/processing technologies. We should test it accordingly.