[Consumer] [Discovery] Automatic retries and dead-letter-queue for event bus

Automatic retries and DLQs (dead letter queues) are often available for different eventing tools.

This does not come as a built-in feature for Kafka (not sure about Redis Streams).

Current workarounds:

1. Replaying idempotent events from the time of the first occurrence of any issue. (Tested with Kafka)
2. Spinning off a celery task that has retry capabilities built-in. (Unused, but possible. Has downside of involving celery.)

A variety of blogs can be found using additional Kafka topics for each retry count and one for DLQ. Kafka partitioning order complicates this. I imagine the DB could also be used, or other options could be explored.

Note that as soon as a consumer needs to make requests to another service, and that service may sometimes be flaky, this may be an issue, because replaying events requires manual intervention by a developer.

No files, tests, or entry points are named. Start by tracing the current Kafka consumer and event-bus flow, then compare retry and dead-letter-queue designs, including partition ordering and possible Redis Streams or database approaches; done requires an agreed design and implementation plan.