DuplicateIDCacheImpl leak

One bridge is created between each node in the cluster. The broker uses a store-and-forward queue for each bridge to move messages between brokers. However, the system is repeatedly creating new store-and-forward queues and the old store-and-forward queues are not cleaned up. This is similar to issue ENTMQBR-2700 for openshift pods that scale down and up.

If we examine the data print, we see lots of bridge queue names, and there should be one queue per bridge connection:

[shiggs@localhost data-print]$ grep "AddRecordTX.BRIDGE." ./broker_data_print.txt | grep -o "address=BRIDGE.*," | sort | uniq | wc -l
79592

It's unclear why all these bridge store and forward queues are being generated.