class HParticleCandFiller: public HReconstructor

 Constructor calls the constructor of class HReconstructor without the name
 and the title as arguments. Optional parameters are explained in setConditions().

 Constructor calls the constructor of class HReconstructor with the name
 and the title as arguments. Optional parameters are explained in setConditions().

 destructor deletes the iterator

 Set parameters by names. Options (may be separated by comma or blank chars),
 By default (no option specified) all values will be kFALSE:
 GOODSEG0   - skip the track when inner seg chi2 < 0
 GOODSEG1   - skip the track when outer seg chi2 < 0
 GOODMETA   - skip the track when no META was fired
 GOODRK     - skip the track when RK chi2 < 0
 KALMAN     - switch from RK to Kalman for mom reco
 GOODLEPTON - skip all candidates which have no RICH matching
 DEBUG      - write out candidate objects for debugging
 NORICHALIGN- dont align rich phi and theta
 NORICHVERTEXCORR- dont do vertex correction in ring sorting
 NOMETAQANORM - dont do normaization of RK dx
 NOGEANTACCEPTANCE - dont fill geant acceptance and crop bits
 NOFAKE     - skip fake marked candidates
 ACCEPTFAKE - do not skip fake marked candidates

 set internal variables to start values

 Get Categories and Parameters etc....

 loop on HMetaMatch2 objects and fill an intermediate
 working candidate object. Selections are performed on this
 objects. All objects which are flagged to be used are
 transported to the output HParticleCand category.

 jobs to be done after last event is finished.

 returns the best rich slot in HMetaMatch2
 corrections for RICH phi+theta are taken into
 account

 create candidate meta objects for all selected matches
 in HMetaMatch2. In Simulation the GEANT information is
 filled too. This function is the heart of the task.
 Selections which META hit to use are performed here.

 for simulation we have to fill HGEANT infos
 in addition. The GEANT track infos are collected
 in trackVec objects for each detector and for
 all detectors together. The tracks in the common
 vector are sorted by 3 criteria:
 1. number of detectors which have seen this track
 2. sum of weights of the track
 3. correlation flag
 The best track number will be transported
 to HParticleCandSim output object.

 From the candidate objects the output HParticleCand
 will be filled if the candidate flage useed == 1
 All additional categories objects will be filled
 in the same order so that the objects later can be
 easily matched.

 clear the working objects at the end of each event.