Large Language Models in Legislative Content Analysis: A Dataset from the Polish Parliament