Toward LLM-Supported Automated Assessment of Critical Thinking Subskills