Numerical analysis of a nonsmooth quasilinear elliptic control problem: I. Explicit second-order optimality conditions