upstream change to superevent_neighbours packet

I have a draft set of changes to how g-events are serialized and queried that's almost ready for testing on gracedb-test. I have a quick sanity check before I do though:

Part of the information in the superevent_neighbours igwn-alert field is the pipeline_preferred_events data for the neighboring superevents. Would it be disruptive to low-latency operations if that field is removed? Is it even used by gwcelery? My understanding was that superevents_neighbours was in there for new event igwn-alerts so you didn't have to re-query for superevents for every new event that gets uploaded, and the pipeline-preferred logic doesn't get populated until after five minutes after the superevent is made anyway.

Removing that field from the superevent_neighbors packet cleans up database querying a lot.

@deep.chatterjee @cody.messick @brandon.piotrzkowski

I'm not sure about other parts of gwcelery, but nothing on the external/RAVEN side uses those fields. I assume @deep.chatterjee would know about the superevent manager.

Just checking, you're asking to remove only the pipeline_preferred_events field and not the entire superevent_neighbors?

yes, only pipeline_preferred_events. everything else in the superevent_neighbors dictionary is the same

Yeah that should not affect anything on our end.

mentioned in merge request computing/gracedb/server!243 (merged)