metadata syncs are not continuous

[sssoleileraaa]

Description

Right now the queues are unpaused when either:

• a user clicks on the retry link
• a MetadataSyncJob succeeds

We currently pause the metadata sync queue whenever a job fails because of a request timeout or api inaccessible error. We don't want to do this. Instead metadata syncs should continuously run and retry in the background until one succeeds, and if the queues are paused, it should unpause them.

[sssoleileraaa]

because of this issue we don't have continuous metadata syncs, and #715 moved metadata sync jobs outside of the main queue, which is a good first step, but we now need to make sure metadata syncs are never paused when there is a failure, instead we should continue to run metadata syncs in the background until one succeeds, in which case we resume the other queue jobs