class AliMUONTrackReconstructorK: public AliMUONVTrackReconstructor

 Constructor

 Destructor

 To make track candidates (assuming linear propagation if AliMUONRecoParam::MakeTrackCandidatesFast() return kTRUE):
 Start with segments station(1..) 4 or 5 then follow track in station 5 or 4.
 Good candidates are made of at least three clusters if both stations are requested (two otherwise).
 Keep only best candidates or all of them according to the flag AliMUONRecoParam::TrackAllTracks().

 To make extra track candidates assuming linear propagation:
 clustering is supposed to be already done
 Start with segments made of 1 cluster in each of the stations 4 and 5 then follow track in remaining chambers.
 Good candidates are made of at least three clusters if both stations are requested (two otherwise).
 Keep only best candidates or all of them according to the flag fgkTrackAllTracks.

 Re-run the kalman filter from the most downstream cluster to the most uptream one
 Return kFALSE in case of failure (i.e. extrapolation problem)

 Re-run the kalman filter from the cluster attached to startingTrackParam to the most uptream cluster
 Return kFALSE in case of failure (i.e. extrapolation problem)

 Follow tracks in stations(1..) 3, 2 and 1

 Follow trackCandidate in chamber(0..) nextChamber and search for compatible cluster(s)
 Keep all possibilities or only the best one(s) according to the flag fgkTrackAllTracks:
 kTRUE:  duplicate "trackCandidate" if there are several possibilities and add the new tracks at the end of
         fRecTracksPtr to avoid conficts with other track candidates at this current stage of the tracking procedure.
         Remove the obsolete "trackCandidate" at the end.
 kFALSE: add only the best cluster(s) to the "trackCandidate". Try to add a couple of clusters in priority.
 return kTRUE if new cluster(s) have been found (otherwise return kFALSE)

 Follow trackCandidate in station(0..) nextStation and search for compatible cluster(s)
 Keep all possibilities or only the best one(s) according to the flag fgkTrackAllTracks:
 kTRUE:  duplicate "trackCandidate" if there are several possibilities and add the new tracks at the end of
         fRecTracksPtr to avoid conficts with other track candidates at this current stage of the tracking procedure.
         Remove the obsolete "trackCandidate" at the end.
 kFALSE: add only the best cluster(s) to the "trackCandidate". Try to add a couple of clusters in priority.
 return kTRUE if new cluster(s) have been found (otherwise return kFALSE)

 Compute new track parameters and their covariances including new cluster using kalman filter
 return the additional track chi2

 Add 1 cluster to the track candidate
 Update chi2 of the track

 Add 2 clusters to the track candidate (order is important)
 Update track and local chi2

 Try to recover the track candidate in the next station
 by removing the worst of the two clusters attached in the current station
 Return kTRUE if recovering succeeds

 Compute new track parameters and their covariances using smoother

 Complete tracks by adding missing clusters (if there is an overlap between
 two detection elements, the track may have two clusters in the same chamber).
 Recompute track parameters and covariances at each clusters.
 Remove tracks getting abnormal (i.e. extrapolation failed...) after being complemented.
 Return kTRUE if one or more tracks have been complemented or removed.

 Improve the given track by removing removable clusters with local chi2 highter than the defined cut
 Removable clusters are identified by the method AliMUONTrack::TagRemovableClusters()
 Recompute track parameters and covariances at the remaining clusters
 and if something goes wrong (i.e. extrapolation failed...) set track chi2 to max value

 Update track parameters and covariances at each attached cluster
 using smoother if required, if not already done
 return kFALSE if the track cannot be extrapolated uo to the last chamber

 re-fit the given track

 Not implemented copy assignment operator