Issue Summary

When the sd.internal.base.db.backed.completion.events.enabled dark feature is enabled, JSM uses the "DB backed" off thread system. The thread pool used to consume events added to PSMQ is named SdSerialisedOffThreadProcessor-n

This pool is fixed to a size of 5, which breaks scaling message consumption when this pool type is in effect.

Customers who have a significant amount of issue events (update/comment etc) could  find that this fixed pool size is not enough to consume events more than they are added, leading to an infinite build up of the AO_319474_MESSAGE table

Steps to Reproduce

1. Set Dark Feature sd.internal.base.db.backed.completion.events.enabled 
2. Put load onto Jira
3. Note that the thread pool is fixed to 5
4. Setting the defined workaround to increase the SdOffThreadEventJobRunner pool size, sd.event.processing.async.thread.pool.count has no effect

Expected Results

It's not clear. If we expect the previous workaround to take effect, the workaround should result in.

If not, then, we should add a new property to define the pool size - preferably, one that does not require DB manipulation.

Actual Results

The pool size is fixed and does not grow

Workaround

It is not possible to increase the SdSerialisedOffThreadProcessor pool size.

The only options are to either:

• Go back to the deadlock risk SdOffThreadEventJobRunner pool, by removing the dark feature internal.base.db.backed.completion.events.enabled
• Disable off-threading completely by setting dark features

  sd.internal.bounded.off.thread.on.completion.events.disabled
  sd.internal.base.off.thread.on.completion.events.disabled