Well most 3D gimbals over light cameras are.
I think it's a non-issue as I see how it works on mine, I will see tilted horizons under strong crosswinds and always in the same tilt angle as the bird.
My theory is that the horizon is mostly corrected from accelerometers values and gravitational pull and the forces resulting from the bird compensating constantly for the wind to stay in place at the birds's scale affect its perception of the vertical by definition gravitational pull, thus tilting the horizon.
Now the Mavic is lighter and the camera being aligned/at the same level with the body and its centre of gravity would mean that the above is even more accentuated in the Mavic's case.
Don't know if any f this makes sense...
Sent from my iPad using
PhantomPilots