Assuming you have a Delta Lake table named customer_purchases that contains duplicate records based on customer_id. You want to deduplicate these records by keeping only the latest purchase record for each customer based on a purchase_date. How would you use the MERGE statement in Databricks SQL to achieve this?