Job runs/enqueued multiple times if server restarts while job is being processed

[rklfss]

Scenario:

• Configure the server to use sliding visibility

  SlidingInvisibilityTimeout = TimeSpan.FromMinutes(5.0),
  QueuePollInterval = TimeSpan.Zero,

• Enqueue a job
• wait for it to be in the processing state
• restart the job server while the job is being processed
• The dashboard shows a notification that the server stopped resopnding
  • click the re-enqueue button in the dashboard for that job before the job is re-enqueued by the system itself
    

Actual result:

• After the sliding visibility timeout is exceeded, a second instance of the job starts.
• That's because the JobQueue table gets another entry for the job and the legacy one is used as well when the FetchedAt timestamp is outrun by the timeout
  

Expected result:

• A job can not run multiple times concurrently
• There is not more than one entry for each job in the JobQueue table, even if the server restarts/crashes etc.