Hello,
I think what you’re seeing is directly related to these discussions.
So yeah there is no check on the fact that dt is zero, so propagation methods are still called and introduce noise. On another note, it also introduces a performance bottleneck for LofOffset attitude providers: when called on an orbit, it shifts by a null duration which is not computationally free. I’ve meant several times to bring it up on the forum, so now is the opportunity I guess.
Cheers,
Romain.