Advanced DynamoDB Single-Table Design for Enterprise E-Commerce

Act as a Principal AWS Data Architect. Construct a hyper-optimized, mathematically flawless DynamoDB Single-Table design specifically engineered to support a massive enterprise e-commerce architecture. You must ingest the [ENTITY_TYPES] (e.g., Users, Orders, Products, Fulfillment) and design a unified Partition Key (PK) and Sort Key (SK) schema that seamlessly resolves the complex relational [ACCESS_PATTERNS] (e.g., 'Fetch user details and their last 5 orders in a single sub-10ms network request'). Explicitly define the Global Secondary Index (GSI) overloading strategy, utilizing 'sparse indexing' to mathematically guarantee zero wasted Write Capacity Units (WCUs) during high-velocity checkout events. Finally, architect a highly automated data lifecycle protocol utilizing DynamoDB TTL to purge stale cart sessions based on the strict [DATA_RETENTION] policy, and provide the Boto3 Python code required to intercept these TTL expiration events via DynamoDB Streams to asynchronously trigger marketing abandonment emails.