View of Fidelity, Robustness, and Plausibility of Post-Hoc Explanations for Diabetic Retinopathy Grading: An Empirical Application of the Clinical-Explainable Artificial Intelligence Evaluation Framework