If a peak exist in the transmission spectrum, then it can only in the most extreme case be a Lorentzian shaped peak with a half-width at half-maximum of the greens function infinitesimal. Therefore if your energy sampling is roughly in the same magnitude with this half-width, then there is no problem.
Futhermore a peak of this type can only exist, if there is a energy band of the electrode, which has a variation in the entire reciprocal space ( k-points variation ) significant less than the greens function infinitesimal.
Therefore if the default value is used for the greens function infinitesimal, then only realistic case were we could still see this problem, would be a perfect 1d system with a surface state, that was a true molecular state, in the band gap, and no scattering at all. In other words, it is not going to happen, unless you try to create a very artificial system to trigger this exact phenomenon.
Perhaps an atomic chain of non-interacting hydrogen atoms.....
