Maximum principle for an optimal control problem associated to a SPDE with nonlinear boundary conditions