[event-hubs] Introduce a random delay when starting the event processor loop

[richardpark-msft]

Add in a randomized delay when starting the event processor loop for the first time, to avoid processors running on the exact same cadence and getting into a deadlock.

Fixes #7017

[richardpark-msft]

/azp run js - eventhub - ci

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[richardpark-msft]

/azp run js - eventhub - ci

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[richardpark-msft]

/azp run js - eventhubs-client - tests

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[richardpark-msft]

/azp run js - eventhubs-client - tests

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[ramya-rao-a]

@chradek, I believe the issue this PR is meant to fix is in the current milestone. Are the changes here still relevant or has anything changed since this PR was created?

@richardpark-msft, can you please update the PR to resolve the merge conflicts?

[chradek]

I might hold off on this until I've gotten the PR for load balancing changes out in case some of how we initialize things changes. .NET is also doing something interesting where they only claim one partition at a time in the loop to prevent these sort of problems, and I'd like to see if we can incorporate some of that in our balanced load balancer strategy.

[richardpark-msft]

It really seems like the need for this doesn't exist. Closing for now, we can always reopen it if we change our minds.