How would you read an execution plan, identify bottlenecks, and refactor a query to reduce runtime for TaskRabbit analytics workloads?
