These blow up the RMSD step for some ways of calculating RMSD. My use of this benchmark isn't totally "inside" - I'm doing the RMSD in its own separate RosettaScripts step - so it might affect only me. I'm filing the issue because I'm having to remove the heteroatoms to get it to work for my purposes. I can PR fixed structures if it is determined that they are wanted, but I'm not sure that it's relevant for the "official" use of the data.