AMONGAGENTS: Evaluating Large Language Models in the Interactive Text-Based Social Deduction Game