Quantifying representativeness in randomized clinical trials using machine learning fairness metrics