References JIRA issue https://issues.redhat.com/projects/DBZ/issues/DBZ-7183 and Github PR https://github.com/debezium/debezium/pull/5036

I only recently had cycles to rollout DBZ 2.5.final to our environment and noticed the millisecond resolution behind source timestamps we contributed to the MySQL connector had a logic bug from a last minute refactoring before release to 2.5 final

This gist of the logic bug is the `eventTimestamp` variable should only evolve on GTID events, not fall through to the seconds resolution fallback - fixed in https://github.com/debezium/debezium/pull/5281