Nt_STM does not really "deal" with the vacuum region between the tip and the sample, because it does not need to.
As a matter of fact, EHM calculates interactions between all atoms of the system (surface+sample+tip). Therefore, when there is no atom, there is no interaction, thus no need for calculation. Indeed, as interactions decrease exponentially with the distance, there is no such thing  as « real » vacuum from this viewpoint, merely interactions that can stretch to an immaterial level.
Hence, as far as Nt_STM is concerned, the code ignores this issue and relies on the exponential decay of the Slater-type AOs to take care of this question in a non-self-consistent fashion.