What are the key differences between Lambda and Kappa Architecture in terms of batch processing, stream processing, and system maintenance? How do their data pipelines handle historical and real-time data? What factors should teams evaluate when selecting an architecture for their analytics requirements?