Using a second‐order differential model to fit data without baselines in protein isothermal chemical denaturation