TaskManager: RabbitMQ outages requires a James restart

[Arsnael]

Steps to reproduce:

• Stop rabbitMQ
• Interact for a while with RabbitMQ
• Restart rabbit.

Observation: RabbitMQ does not automatically heal!

A James restartis being required.

2020-11-24 05:01:03.164 [error] <0.758.0> Channel error on connection <0.730.0> (192.168.0.5:55846 -> 192.168.0.3:5672, vhost: '/', user: 'guest'), channel 1:
operation basic.consume caused a channel exception not_found: no queue 'eventbus-b0ce1474-c6e5-4117-a611-2a62d1e11ec3' in vhost '/'
2020-11-24 05:01:03.192 [error] <0.854.0> Channel error on connection <0.730.0> (192.168.0.5:55846 -> 192.168.0.3:5672, vhost: '/', user: 'guest'), channel 14:
operation basic.consume caused a channel exception not_found: no queue 'terminationSubscriber6557b9f9-812c-42e8-9020-0b6646731398' in vhost '/'
2020-11-24 05:01:03.200 [error] <0.862.0> Channel error on connection <0.730.0> (192.168.0.5:55846 -> 192.168.0.3:5672, vhost: '/', user: 'guest'), channel 16:
operation basic.consume caused a channel exception not_found: no queue 'taskManagerCancelRequestsQueue89eddcec-542c-4237-b379-656a692af78e' in vhost '/'
2020-11-24 05:01:03.426 [info] <0.910.0> accepting AMQP connection <0.910.0> (192.168.0.5:55848 -> 192.168.0.3:5672)

This, is due to the use of exclusive queue, that do not gets re-created upon restoration of the connection.

[chibenwa]

Maybe we should have a "Recovery callback" should be implemented as part of SimpleConnectionPool.

We can use this callback to recreate the queues when we restore a session.

Work on this should be handled as a draft. QA should be performed on DEV before a merge.