[xCluster DDL Replication] ddl_queue should not report SYSTEM_ERROR during regular replication

Jira Link: DB-23584 Today ddl_queue reports a system_error when waiting for other streams to catch up to xcluster safe time. This causes extra errors on YBA for each DDL that is run, which causes extra noise. We should not report this as an error since it is expected during regular replication

We should perhaps also include a specific error message for when ddl_queue is actually paused, and also see if we can bubble this up to the master as well