Confidence Under the Hood: An Investigation into the Confidence-Probability Alignment in Large Language Models