RabbitMQ queues are not declared as replicated

[cailyoung]

We noticed during a RabbitMQ node outage recently that log_message_saving was unavailable to the jobs service despite 2/3 of the RabbitMQ nodes being up. After investigation we noticed that (all?) of the queues declared by this and other services are 'Classic' queues, which exist only on whichever node the client that declared them was connected to at declaration time.

It should be considered to allow replicated queues, either via quorum queues or streams, so that multi-node RabbitMQ systems can be operated safely and not cause consumer failures for ReportPortal.

[DzmitryHumianiuk]

@cailyoung definitely makes sense

[DzmitryHumianiuk]

related: https://github.com/reportportal/service-api/issues/1744 https://github.com/reportportal/service-jobs/issues/86 https://github.com/reportportal/service-api/issues/1745