noctuid
I previously removed incremental queuing to ensure consistent final documents (and for other reasons). For extremely large sessions, the current parsing could take a long time. I should first try to address any low-hanging fruit for parsing speed and then consider going back to appending or keep replace but not with a full rebuild.