Reduce DB contention on Ingest

At the moment we insert large batches of events that go and manipulate lots of tables. In practice when ingesting tons of events this causes too much contention which causes ingest to fail. Switch to doing ingestion across several transactions to reduce contention. This is a bit slower but does not fall on it's face when lots of concurrent ingestion is happening (from the cross instance event synchronizer for example). I might want to later update again how this works to be a bit more efficient but this works for now I think.