kcidb_spool_notifications times out often

The `kcidb_spool_notifications` Cloud Function has been timing out a lot recently:

We need to improve performance and reduce load in order to let the stuck updates through, and prevent congestion in the future.

I'll reduce the number of objects posted per message into the update message queue, and will try to disable python assertions in cloud functions as a mitigation.

Start by locating the kcidb_spool_notifications Cloud Function and reviewing how it posts objects into the update message queue. Use the reported timeout behavior and the proposed reduction in objects per message as the starting point; done means the function no longer frequently times out and stuck updates can pass through without congestion.

Written by the indexing model from the issue text.