For the extension to recognize that source files belong to a database's source archive, we currently require that the source archived is a .zip archive, as opposed to a src/ directory. This requirement is satisfied already by databases created with the cli tools, but LGTM still produces databases with odasa with unpacked src/ directories. So the experience the user has when they download a database from lgtm.com is that they have to run codeql database cleanup to zip the source archive before, e.g., jump-to-definition will work.

Some options available are:

1. Simply provide a convenience command in the extension to run cleanup on a given database.
2. Prompt the user to ask whether they want to run cleanup (though possibly with the question phrased in a more perspicuous way) when a database with unpacked src/ is added to the workspace.
3. Prompt the user to ask whether they want to run cleanup when running a query.

We almost surely don't want to silently do the cleanup --- we have deliberately avoided doing any other in-place changes to user databases without notification or offering the ability to decline it.