When Your AI Agent Succumbs to Peer-Pressure: Studying Opinion-Change Dynamics of LLMs