CLICK_STREAM in BigQuery stores epoch times in column DT but all fields were loaded as STRING. You need DT as TIMESTAMP with minimal migration effort and low future query cost. What should you do?

Creating a new table with the correct schema and migrating data is the most efficient solution for large datasets in BigQuery. It ensures DT is stored as TIMESTAMP directly, optimizing future query performance and storage costs. Dropping and recreating the table would require a full data reload, which can be time-consuming and disruptive. Adding a new TIMESTAMP column to the existing table still leaves the original STRING column, potentially leading to confusion and not fully optimizing storage. Creating a view that casts DT to TIMESTAMP would incur casting costs on every query, increasing query time and cost. The option to add TS and ISNEW and reload in append mode is overly complex and doesn't fully address the underlying schema issue for existing data.