Currently, we configure Camus jobs to import events based on which eventgate cluster instance they come through. We do this because generally the throughput of each stream is relatively the same size for each eventgate cluster instance, which allows us to better tune the ingestion jobs.
T269619: Create pipelines for late/spurious/failed events will create one of the first set of event platform based streams that does not flow through eventgate. We need to add a Camus job that knows how to import these kinda of events, probably by adding another stream config setting to indicate that they should be ingested by Camus. This may cause us to re-think which stream config setting we are using to configure ingestion altogether.