after the discussion of last SeeVogh meeting, and after several tests, I realized the better solution to avoid the stops in the geane propagation for long distances and to create a new release in a very short time is to modify the Geane.C, proposing larger epsilon for the "dangerous" materials:
m->SetParam(6, 0.005); // epsil
m2->SetParam(6, 0.005); // epsil
m3->SetParam(6, 0.005); // epsil
This modification is in the trunk since yesterday evening.