In the context of Spark Structured Streaming, you are tasked with implementing a solution that requires maintaining and updating the state of data over time to perform time-based aggregations, such as windowed averages or counts. The solution must be scalable, handle large volumes of data efficiently, and comply with the organization's cost constraints. Which of the following options BEST describes the correct approach to implement stateful processing in Spark Structured Streaming for this scenario? Choose the single best option.