As far as I can tell, that supports, rather than contradicts, the post's point about modelling, fp, and approximation errors. It's a little too easy to ascribe KSP's bugs to floating-point errors, its physics engine just isn't trustworthy enough. How do you really know that the oscillations are not just due to it solving a stiff system of ODEs with explicit Euler, or something perfectly predictable like that?

Thus the simulation resembles reality, where such oscillations arise out of natural imperfections ;)