Exploiting Differential Flatness and Pseudospectral Optimization to Improve the Computational Efficiency of Health-Conscious Lithium-Ion Battery Control