What Does the AI Doctor Value? Auditing Pluralism in the Clinical Ethics of Language Models — ThinkLLM