How would you handle slow query performance for a single-user SQL endpoint in Databricks, where all sequentially run queries are affected?