bug: handle "sqlite3.OperationalError: database is locked" error

[terriko]

Our test suite has been sometimes throwing the following error:

sqlite3.OperationalError: database is locked

I think some of our immediate problem can be fixed with improving our test design as I described in #4651 but I'd also like to have this error caught and handled appropriately. It may make sense to have this be a "back off and retry" loop but since the database updates can take 10-20 minutes (maybe even longer if you're talking NVD API on a slow internet connection?) we might also consider just failing more gracefully and directing people to the documentation on running concurrent scans. Or spitting out an actual error message before we enter the retry loop?