Stationarity conditions and scalarization in multiobjective optimal control of a nonsmooth PDE