[Core] Explanation of the --empty flag to support unit tests

[dbeatty10]

Link to the page(s) on docs.getdbt.com requiring updates

• https://docs.getdbt.com/reference/commands/run#the---empty-flag
• https://docs.getdbt.com/reference/commands/build#the---empty-flag
• https://docs.getdbt.com/docs/build/unit-tests#:~:text=This%20enables%20you%20to%20write%20succinct%20and%C2%A0specific%C2%A0unit%20tests.

Tell us more about this update

This content feels like it could be combined, simplified, and unified (potentially into a shared snippet).

As-is the pages for run and build are missing the same level of detail that is contained in the unit tests page.

Simultaneously, the unit tests page feels like it might have too much detail.

Reviewers/Stakeholders/SMEs

.

Related GitHub issues

dbt-labs/dbt-core#10695

Additional information

💡 The key insight

When you run dbt tests in a stand-alone fashion (e.g., dbt test), it needs the database tables, views, etc. to exist in the database! Otherwise, it will give an error like "relation doesn't exist" 💥

👉 So doing a dbt run --empty creates all the database objects in the dbt project in the most minimal way possible -- just the column headers + data types, but no rows. Building these objects with zero rows is faster and incurs less warehouse spend than building the entire data set.

✅ Then after that, you're free to do dbt test --select some_specific_test as much as you want!