StreetviewLLM: Extracting Geographic Information Using a Chain-of-Thought Multimodal Large Language Model