Handle the cycle in queue manager.

[style95]

Steps to reproduce the issue:

It's not easy to reproduce it as it occasionally happens.
But what we observed is as follows.

1. Queue endpoints are suddenly removed at once for some reason.
2. The DataMangementService tried to put the queue endpoint again but there was already data written.
3. So it just removed the queue as it thought there is another queue running.
4. But there was no queue running and the queue endpoint was the same scheduler.
5. So there was no queue but queue endpoint data was in ETCD.
6. Since the endpoint in ETCD points to the scheduler without a queue, activations were sent to the scheduler repeatedly.

Additional information you deem important:

• Need to test when etcd data is abnormally removed and recovered.

[bdoyle0182]

I think I just hit this for the first time for one queue. If we don't have a fix for this, it would be great if we could include the remediation steps even if it's just a restart of one of the components. To me, it seemed like restarting both the scheduler and controller was needed