fix: Fix metadata reflection without a default dataset

[JacobHayes]

Fixes #838 and fixes #1088. 🦕

The get_table_names and get_view_names methods are supposed to return the bare names (no {schema}. prefix) of the resources for a single schema/dataset (where schema=None is the "default schema", not "all schemas"). However, the BigQueryDialect implementation returns:

• bare names for a specific schema: if the connection has a default dataset
• {schema}. prefixed names for one schema: if the connection doesn't have a default dataset and the schema arg is a string
• {schema}. prefixed names for all schemas: if the connection doesn't have a default dataset and the schema arg is None

The bolded parts cause issues outlined in #1088. This PR fixes the get_table_names and get_view_names implementations to return bare names as SQLAlchemy expects.

This is a breaking change for a subset of users:

• (no impact) connections with a default dataset (likely a majority of users?) should behave the same
• (impact) connections without a default dataset using Metadata.reflect() without a schema argument were probably expecting to reflect all datasets, but they would now not reflect anything.
  • BigQuery doesn't have the same concept of a default schema the way eg: PostgreSQL does (public) - so without a default set on the connection string, we don't have anything to search and return an empty list.
• (no impact) connections without a default dataset using Alembic were probably broken anyway

This has been working for me locally with alembic + no default dataset (the models specify the schema) + include_schemas=True (required for cross-schema stuff).

---

• Make sure to open an issue as a bug/issue before writing your code! That way we can discuss the change, evaluate designs, and agree on the general idea
• Ensure the tests and linter pass
  • unit tests + linter pass locally
• Code coverage does not decrease (if any source code was changed)
• Appropriate docs were updated (if necessary)

[conventional-commit-lint-gcf]

🤖 I detect that the PR title and the commit message differ and there's only one commit. To use the PR title for the commit history, you can use Github's automerge feature with squashing, or use automerge label. Good luck human!

-- conventional-commit-lint bot
https://conventionalcommits.org/

[JacobHayes]

Hi, any chance I could get the tests approved to run on the PR? I've been trying to setup the system tests locally, but they're a bit onerous (understandably).

[Linchin]

Sorry for the delay, just started running the tests!

[JacobHayes]

Sorry for the delay, just started running the tests!

no worries, thank you!

[chalmerlowe]

@lingchin if this is a breaking change as Jacob indicates, then we need to have some sort of deprecation warning that the break is coming and get this into the schedule.

This is a breaking change for a subset of users:

• (no impact) connections with a default dataset (likely a majority of users?) should behave the same
• (impact) connections without a default dataset using Metadata.reflect() without a schema argument were probably expecting to reflect all datasets, but they would now not reflect anything. BigQuery doesn't have the same concept of a default schema the way eg: PostgreSQL does (public) - so without a default set on the connection string, we don't have anything to search and return an empty list.
• (no impact) connections without a default dataset using Alembic were probably broken anyway