Clippy job in the pipeline can use sqlx Offline mode

[9SMTM6]

The clippy job, unless I missed something or did not get there yet, (currently chapter 7) doesn't require to use a live database. So it can be shortend (importantly also in execution time and resource usage) by removing everything but the checkout, the cache and then the linting task.

Its generally somewhat annoying to me how many tasks are repeated between the different jobs, but admittedly I also cant find an approach that does this better without tradeoffs (such as doing them only as steps in one task, or committing some image to an image repository, which requires more buy in into some container image repository such as github, and might leave behind a lot of garbage).