The resolution to this paradox is realizing that the electrons are originating from two *different initial states*, one photo-emitted from surface A, $\vert \psi_A\rangle$, and the other surface B, $\vert \psi_B\rangle$. Because the initial states for your two scenarios are different, they naturally have different binding energies (i.e. different starting potential energies) so you expect they will have different kinetic energies after photoemission as well. Thus, there is no violation of energy conservation.