Always set application_name in primary_conninfo
With v11 and before, application name was not falling back to the cluster name, leaving it to the default 'walreceiver'.
Because of this, wait_for_catchup fails when waiting for a specific standby on a primary having multiple ones. The function waits for a single row, but get multiple ones because all standbys match with filter "application_name in (..., 'walreceiver').