It's not a matter of "round enough", as if roundness were somehow an important feature. It is the internal physics-- what force is balancing gravity? If the answer is molecular bonds, which are capable of exerting forces in arbitrary directions, then roundness need not appear-- any more than it does in a common boulder. If the answer is isotropic pressure, then you must reach the equipotential shape because such pressure forces can only exert forces perpendicular to the surface. So "roundness" is simply a shortcut for identifying the internal force balance, a crucial element for understanding the internal physics. I wouldn't be against instead using differentiation as the key internal physics issue, and that might limit us to somewhat larger objects than just the roundness constraint. For one thing, it rules out "piles of dust" that are weakly gravitationally bound.