The study you refer to seem to assume that if we find a even distribution of matter it can't be a fractal. Where is the support for that definition of a fractal. To me a fractal pack it self up into a (self) likeness of itself, at all scales measurable? But it does not state that there need to be a randomness to it? If we assume that a fractal is random we also lose the properties we expect making fractal behavior interesting for explaining how the very small can 'compress' information of the very big. And you have a point in saying that we can't guarantee what the non visible part of the universe might look as. But the homogeneity and isotropy of space is assumed to exist at all levels, just as we assume that all physics experiments will behave the same even outside our observational field. And as that proposition is more likely built on what so far have learned about the universe we can see their point there, can't we?

So the question, to me, becomes if there is a demand for a fractal not to, ever, be evenly distributed at some scale, although uneven in some other for example? If that is a mathematical fact then they are possibly correct, but if you can have 'layers on layers' mathematically in where some present a homogeneity whilst other permutations differ from that then? They should need to be more careful in their statement.