Remarks on the use and misuse of the semi-inverse method in the nonlinear theory of elasticity