Mark updated dataset rows for timely model retrains

Your dataset rows arrive without reliable update timestamps, causing stale feature joins and delayed model retrains. Provide consistent timestamps and ready-for-retrain flags so retraining runs before the next scheduled job.