Exploring the Potential of Large Language Models for Estimating the Reading Comprehension Question Difficulty