I’ve just had a run-in with this bug myself.
Why are we still displaying higher-order IDs rather than the leading ID of an infraspecies, or at least the parent species, if there is a problem with the way RG is coded? If I apply an ID for an infraspecies, I am also implicitly applying an ID of the species. In the case of the autonym, I am adding an ID for the species explicitly.
If the problem is with the way we handle RG, then we should be turning off RG until we find a way to solve this, not preventing identifications of infraspecies. I find the current concept of Research Grade a bit of a joke currently anyway. It is constantly gamed by a relatively high proportion of users blindly and instantly agreeing with external IDs.