Contextualized Streaming End-to-End Speech Recognition with Trie-Based Deep Biasing and Shallow Fusion