I think the idea of the restriction is to close the loophole where you can, in principle, create a species for any environment by repeated applications of low-level gene mods. At some point, it would become more economical to simply mod a species twice than to research an expensive next level of genetics tech. If you already have, for example, +10%, +20%, and +30% temperature mods, it is probably better to mod your base species to +30%, then mod that species by another +30% instead of researching three techs of increasing cost (40%, 50%, 60%) to get the same effect.
Personally, I don't see the need to "cull" out-of-date species. If a species was good enough to settle a particular colony, you can just... leave them there? Let that population grow and then move them to another body with similar characteristics? I think the mistake here is the assumption that you (1) must invent new species whenever you have new genetics tech, and/or (2) must use the latest/highest level of genetics tech exclusively. Rather, I think the idea is that we should design a species for a specific purpose, i.e., to colonize a specific body or set of bodies. From that perspective, there is not really such a thing as an "obsolete species", merely species suitable for different missions.
That said, I would like to see something done here as it is an arbitrary restriction and that means we can do better somehow. One idea is that genetic modification techs should apply to single species - this makes sense, if we research the human genetic code and devise a way to give humans +20% gravity tolerance, it makes sense that we would need to re-learn the genetic coding of the high-gravity human species in order to give them (an additional) +20% gravity tolerance. Maybe there is something there as long as the balance of research costs is close enough that both options are viable so a decision is created for the player.
