Now using the custom propagator in the detector simulation: one step forward to solve issue #1 (closed) .
Removed the forward declaration of TFRPropagator and TFRGeometry: is it really needed to have TFRPropagator as member of TFRGeometry? Cross-inclusions and linking (as well forward declarations) must be used only in really specific cases, since they are not optimal.
Setting of the tracks in the detector_sim.