I don't buy this energy argument. I give you a cube and a trillion charges. Are you saying that to minimize the electrostatic energy you would place them all on the surface because then the distance between the charges is smaller? That doesn't seem to be plausible at all.
Let me give an argument without any complicated math. In a metal, you have a reservoir of free electrons causing any added charge (that could be immobile) to be screened. You can simply treat the free electrons as a plasma and then consider this as Debye screening with a very short Debye length. The inverse Debye length can be considered to be the effective mass of the photon, which is thus quite large. You get a charge on the surface because the screening charges have to come from somewhere.
First consider very simple model for a massive photon. Suppose that neutrinos are millicharged particles and then the neutrino backround (which is neutral on average) will lead to a Debye screening of charges so we have an effective photon mass.
If we consider again the charge in the conductor, then the neutrinos do part of the screening, so the free electrons will screen the charge slightly less. Of course, this leads to the same result: The net charge in the bulk is zero and you only have a surface charge.
Now, since the neutrino background surrounding the charge has a net charge (note that the Debye screening length due to the neutrinos is huge), the metal will have the opposite charge distribution due to the electrons so that the total charge will be zero in the metal.
But now we have to consider that the charge distribution in the metal due to the free electrons is the way it is, simply because the way the neutrino screened electric field of the charge depends on the distance from the charge, not because of whether or not the charge is really screened by neutrinos or not.
So, if the photon has a mass which is not due to a millicharged background, then you will inevitably get a net charge in the bulk of the metal.