Enhancing Interpretable Image Classification Through LLM Agents and Conditional Concept Bottleneck Models