Extracting Patient History from Clinical Text: A Comparative Study of Clinical Large Language Models