A failed connection to the internal DB (e.g. during DB shutdown) could result in an increase of failed_count

• Ensure failed db calls result in an invalid and not failed metric
• Wait until all running checks have completed (or have been cancelled) before initiating the shutdown of the DB
  • For checks that require a non-trivial implementation to support cancellation, log backlogged bugs