Adding this extra charge epsilon means we change the total charge of the system. In the original system the total charge is zero, but in model A it is -epsilon. We could solve this by adding epsilon/2 to each electron. This would mean that the constant C in our E_0 would change (it will grow: e -> e+eps/2 => e^2/(4*pi*eps0) -> (e+eps/2)*e/(4*pi*eps0)). This will result in less attraction.