This is not implemented in LAMMPS, but wouldn’t be hard to do. You would just add the correction energy, similar to the self-energy correction. A somewhat related correction (dipole in the z-direction) is already implemented for the slab Ewald.
Yeah, just to confirm that it's easily implemented, I did this in an earlier version a few years ago. You can do it in about 10 lines of code; even I could figure it out! Not sure why it's not a standard option by now, it could be useful for some systems with large dipole moments, large polarization etc. Maybe it's just a matter of not wanting to make Ewald sums even more conceptually difficult; we'd have a zillion more questions showing up here every day about which boundary conditions to use, etc.