Investigate mismatch between model damping and fmi damping using ODE in damped_pendulum example

[traversaro]

If one launches the example in https://github.com/robotology/gazebo-fmi/tree/master/plugins/actuator/examples/damped_pendulum in ODE instead of using Simbody as documented in the documentation, the evolution of the two pendulum are quite different, with the FMI one that seems to be much less "damped" w.r.t. to the "native" one. On Simbody, on the other hand, the evolution is extremely similar. The difference is quite big, so it may be suggest that there is something strange either in the damping implementation in ODE, or in the joint velocity feedback.

This would be something that would be interesting to analyses, probably by adding a specific test comparing the high-precision solution of a damped pendulum model obtained by integrating the model with a variable step integrator with really low error tolerance.