kalman.cpp (Bug #4439)

You kalman.cpp sample (not module) seems to compare wrong entities. In particular, you compare measurement with prediction while
you have to compare measurement with estimation. In other word, instead of using prediction = KF.statePre use KF.statePost that you
have to update BEFORE drawing lines. In the current code, your comparison isn't indicative of Kalman Filter efficiency and only reflects
the fact that during the initialization of the filter you set process noise level (1e-5) to be much smaller value than measurement noise (1e-1).

Your help() function, however, describes the situation correctly: "The real and the estimated points are connected with yellow line segment”
so your intentions were right but the implementation was wrong.