explain the principles and best practices of DynamoDB capacity planning and auto-scaling, including techniques for monitoring and optimizing throughput utilization, managing capacity provisioning costs, and dynamically adjusting capacity based on workload patterns and performance metrics?