Use factory_boy for catalog test data generation

[sarayourfriend]

Problem

The API uses factory_boy for generating test data. The catalog tests do not, and as such, require far more manual configuration of test data, for example:

https://github.com/WordPress/openverse/blob/change/tag-upsert-strategy/catalog/tests/dags/database/test_batched_update.py#L77-L106

This code is in a private function in one test suite, and defaults must manually be configured instead of automatically handled.

This increases the overhead of testing with sample data, which makes writing tests less pleasant over time.

Description

factory_boy has bulk creation methods, and other useful approaches for making the generation of test data a non-issue when writing tests, and centralises the implementation of test data generation so improvements for one test can be shared by all existing and future tests.

We can use factory_boy to generate test data, but inserting it into the database will still require translating the class instance data, because while it supports several ORMs, our Airflow DAGs do not use any of them, not even for table description.

Once we implement #416 we can use the SQLAlchemy (or similar) integrations available for factory_boy, so that test data is easy to generate and maintain.

In the mean time, factory_boy can be used to generate instances of DTOs that get translated to the dicts used to insert data into one table or another as needed, using our own methods to translate the DTOs into those dicts.

Alternatives

Implement our own version of this with bespoke functions similar to the one in the code I linked above. That could be fine too, it just doesn't come with the benefits of years of maintenance and experience in a library like factory_boy meant to make this specific thing easy.