Streaming Large Datasets with RSC and Suspense

Act as a Big Data Frontend Architect. Architect a highly performant streaming solution designed to pipe a massive [DATASET_SIZE] database record set directly to the browser utilizing React Server Components (RSC) and Suspense. Implement a programmatic [CHUNKING_STRATEGY] leveraging async generators or pagination cursors to incrementally yield data chunks to the client without blowing up V8 memory limits. Design an elegant [LOADING_UI_PATTERN] utilizing nested `<Suspense fallback={...}>` boundaries and skeleton screens to mathematically guarantee the DOM layout remains perfectly stable without Cumulative Layout Shift (CLS) while the HTTP stream remains open.