Gets the primal infeasibility measure. If primal feasibility is obtained, the value 0
is returned. Otherwise, this method returns a positive value. Since the network optimizer sees only the embedded network part of the full model, a solution considered to be feasible for the network optimizer may not be so for the full model.