The issue is that "sqrt(arbitrary quartic)" is already more specialized than "elliptic integral". So we can't just talk about elliptic integrals in general, we need proof that this specialization doesn't give rise to closed forms.