State-constrained optimal control governed by non-well-posed parabolic differential equations