Nagios check_queue probe should report unknown instead of critical when the data file is not populated

[Chris Glass]

The check_queue nagios check reports "Critical" when the data file it uses to check on queue lenght is not yet populated (it's a race condition - it sometimes takes a few minutes until the file is populated).

Note that the file is present in this case, just doesn't contain any information yet, triggering the condition at https://bazaar.launchpad.net/~charmers/charms/precise/rabbitmq-server/trunk/view/head:/scripts/check_rabbitmq_queues.py#L52

Nagios documentation specifies that absence of low-level information should trigger a (soft) alert as unknown, since that will automatically recover when the information is made available (or escalate to a hard alert once the user-defined threshold is reached) https://assets.nagios.com/downloads/nagioscore/docs/nagioscore/3/en/statetypes.html