Fix holding of scheduler global lock while looping queues

Remove the looping through the retrieve queue jobs while holding the global lock in OStoreDB::fetchMountInfo.
This was done when a tape was disabled, since we mount disabled tapes if they have pending retrieve requests.
Instead we now mount the tape if the queue object info as in the mount policy map a mount policy whose name
starts with "repack".

With this change, cta-admin now forces repacks added with the --disabledtape to have a mount policy
prefixed with repack (without the --disabledtape flag, the mount policy can have any name)

Also Removed the criteria that a disabled/broken tape is mounted if the queue is filled with deleted
retrieved jobs. Eventually the maxrequestage for the queue will expire, the tape will be mounted
and the queue flushed