I think it is more parsimonious to say that there are no intrinsic properties of the world that are not relational, not that they are impossible to separate from observation. It defeats the purpose of interpretations like relational quantum mechanics to hold on to the concept of an intrinsic reality when the interpretation makes such a concept superfluous.
I would say that it is impossible to disentangle the intrinsic properties of the world from observations in the same way that going north of the North Pole is impossible.
Or to put it another way: observers are intrinsic to reality as well as the systems under observation, and any attempt to treat them separately will fail. But this is uninteresting, because any interpretation is going to have to acknowledge this at some level (as Bohr correctly pointed out quite a long time ago.)
Input 1: authors check out (PRL, Nature pubs). Not a one-off crazy posting to arXiv.
Input 2: Unless this your field, I'd be surprised if you get much from the paper other than interpretation of quantum mechanics is still an active field and information theory / thermodynamics has an important role in that field.
Input 3: After a few reads, I don't like the phrasing the authors put on Type-II interpretations. They say "Another class sees quantum probabilities as not directly dealing with intrinsic properties of the world but with relational experiences between an observer and the world." Another way to say it may be that it's impossible to separate observation from the intrinsic properties of the world. That is, reality is tightly bound up with observation and trying to separate the two is artificial. But I'd be happy to hear what others think about it.