Update eval_gpt_review.py

[mzamini92]

some optimization on the code. set the formatter_class to argparse.ArgumentDefaultsHelpFormatter. This will include the default values of the command-line arguments in the help message, making it easier for users to understand the default behavior of the script. Instead of using question_idx_list and total_len to iterate over the question indices, directly use enumerate in the for loop.