How can you determine if a large table in Snowflake would benefit from defining an explicit clustering key?