When doing the initial schema discovery for a database with 1000+ tables, the postgres replication connection times out.
This happens because we create the replication connection and then do all the work to get catalog information. If this process takes more than a minute, then the postgres backend kills the replication connection due to a timeout.
The jdbc API relies on us to either call `read` or to `forceStatusUpdate` in order to have the statusUpdateInterval to cause a keep alive message to be sent to the server.
I have a potential fix for this that relies on creating a temporary background thread