How do you implement incremental data loading in an ETL pipeline to avoid reprocessing historical data?
