The problem with non-spherical stars or planets is that it's hard to find a justification for them, unless you're willing to throw out conventional physics.
But if you're going to throw out conventional physics, might as well go all the way and reinvent physics from scratch. Forget about planets and orbits, create a whole new universe from first principles. The result will probably look nothing like anything we know or understand, but it would work.
(In fact, I've been working on and off on an alternative universe in n dimensions that's not based on physics as we know it, but based on symmetry.)
Or just invent a fantasy world where we declare things that we want to exist by fiat, and call it a day. I mean, after all, if we're already not sticking with the consequences of physics generalized to 4D, why tie our hands behind our backs?
The whole point behind generalizing 3D physics to 4D is to use things familiar to us to explore 4D geometry. For that purpose, it doesn't really matter if 4D orbits aren't actually stable; we just declare it so by fiat and use that to explore 4D geometry. But if we're going to create a 4D universe that's native and self-consistent, then we can't get away from grappling with fundamental issues like these, at which point the question becomes, do we want to stick with the dimensional analogue of 3D physics and be bound to the consequences which don't always go our way, or should we just throw out the whole thing and invent a 4D universe from first principles?
Alternatively, if we're going to stick with the dimensional analogue of 3D physics, another approach could be to ask the question, if orbits and atoms are not stable, then what
is stable? This will probably lead to a completely new, alien world that's totally unfamiliar to us, but it will also be an interesting exploration of the consequences of physics as we know it applied to a 4D world.