PROPER: A Progressive Learning Framework for Personalized Large Language Models with Group-Level Adaptation