Randomize outbox cleanup sleep duration and/or add jitter to decrease running cleanup on multiple instances concurrently

[ramonsmits]

Describe the suggested improvement

Currently, the sleep interval for outbox cleanup is fixed. A user reported that often this causes issue just after a deployment of a new version as all instances are pretty much aligned on the sleep duration.

To minimize overlap in cleanup it could

• Have a random delay on the first outbox cleanup (for example rnd(1)*CleanupInterval)
• Add jitter to each sleep to decrease the change of running cleanup on multiple instances (for example CleanupInterval/2 + rnd(1)CleanUpInterval)

Additional Context

No response