These are good points, but I would just add that boundaries do not need to be sharp (or in a single descriptive category) for there to be a real distinction that merits categorization.

I'm not sure this actually changes your point, but it raises the possibility of compatibility between what zimu is calling a realist position and a nominalist position, because the nominalist position could just be accepting that we have trouble talking about the real categories precisely and so we do the best we can with the understanding we have and the language we have. (Philosophically there is a difference in kind between realism and nominalism, but I don't think that matters for the purposes of this discussion.)