We ingest data incrementally from a database into delta tables using a column updatedUtc. This column is a datetime and is updated when the row in the database table changes. What about using this non-mutable column in "cluster by"? Would it require executing OPTIMIZE more often?