Problem: we sometimes create a new consensus tree when we shouldn't
Sometimes when reproducing the current state of nostrocket, it publishes a new consensus tree when one exists already.
It seems to be missing an event or something.
Solution:
A quick and dirty solution would be to check the timestamp when attempting to process a state change event out of consensus mode (the only time when we need to produce a new consensus tree is when seeing a valid state change event that is not currently included in the consensus tree).
If the timestamp of the event is not recent, then we are probably missing a consensus event somewhere and should try a bit harder to find it before creating a new consensus tree.