Meltano

I was working on tap-bigquery following closely tap-athena screencast. And I have couple of suggestions for SQLConnector design:

Currently method `discover_catalog_entries` relies on sqlalchemy introspection working correctly and does not support customization.

In bigquery case, due to non-straightforward logic sqlalchemy-bigquery does not work as expected

(in tables discovery it returns table name with schema e.g `schema1.table`, and in getting table properties when schema is also provided as separate argument, treats schema in table name as project name -&gt; fails to find such table)

I think it would be nice to split `discover_catalog_entries` into several overridable functions:

• `_get_all_schemas`
• `_get_all_tables_in_schema`
• `_get_all_columns_in_table`
in this way I would be able to override just a `get_all_tables` part and reuse the rest of the logic

Hi <@U06BX5HMGNB>, we're all for better API design! Would you be willing to contribute an MR for this refactor?

Awesome! Feel free to tag me and ask questions :slightly_smiling_face:

<https://gitlab.com/meltano/sdk/-/merge_requests/305>

I'm not sure about extracted functions names, feel free to suggest alternatives