Convert some arrays that were fixed length to vectors.
  Since these arrays are only used local to the method
  ThcHallCSpectrometer::TrackCalc(), they were moved out of the class
  into the method.
  The affected arrays are fX2D, fY2D, f2XHits and f2YHits

    Structure TOFPInfo holds information for hits within a plane
    Structure TOFCalc holds hit information for all the hits.
    fdEdX changed from 2d array to vector of vectors

      Selection of golden track using prune method is added.
      A bug is also fixed in THcHodoscope class
      Number of pmts hits, focal plane time, good time for plane 4
      and good time for plane 3 are set to the tracks in
      THcHodoscope class.

  This insures that 5th exponent is zero if not specified on recon file

  This is anticipation of a new hit class for processed hits.
  To make it clear that this hit class is for decoding into hit lists

    Add fdbg_raw_cal flag in THcShower.h.
    Initiate fdbg_raw_cal with engine's hdbg_raw_cal in THcShower::ReadDataBase.
    Supply calorimeter debug flags to otherwise free print-outs in the
    THcShower and THcShowerPlane methods.

    Add fdbg_init_cal flag in THcShower.h.
    Initialize fdbg_init_cal from engine's hdebugflaggeometry
    in THcShower::ReadDataBase.

debug info should be suppressed in the main develop branch

- no need to call Nint() on an integer

  Use InitArrays method in THcCherenkov and THcAerogel
  Null out all array pointers in the InitArrays method and
  call that method from all constructors so that it is always
  safe to delete these arrays.

Found when doing valgrind --tool=memcheck --track-origins=yes --leak-check=full hcana
Return errors like below
==14467== Invalid write of size 1
==14467==    at 0x4A0794F: __GI_strcpy (mc_replace_strmem.c:313)
==14467==    by 0x57137A3: THcHodoscope::Setup(char const*, char const*) (THcHodoscope.cxx:112)
==14467==    by 0x5713A7E: THcHodoscope::Init(TDatime const&) (THcHodoscope.cxx:142)
==14467==    by 0x4D38C29: THaApparatus::Init(TDatime const&) (THaApparatus.C:193)
==14467==    by 0x4D67E57: THaAnalyzer::InitModules(TList const*, TDatime&, int, char const*) (THaAnalyzer.C:433)
==14467==    by 0x4D68B1F: THaAnalyzer::DoInit(THaRunBase*) (THaAnalyzer.C:667)
==14467==    by 0x4D68126: THaAnalyzer::Init(THaRunBase*) (THaAnalyzer.C:467)
==14467==    by 0x4D6B31A: THaAnalyzer::Process(THaRunBase*) (THaAnalyzer.C:1289)
==14467==    by 0x4E21E97: G__haDict_704_0_6(G__value*, char const*, G__param*, int) (haDict.C:23491)
==14467==    by 0x6DB6148: Cint::G__ExceptionWrapper(int (*)(G__value*, char const*, G__param*, int), G__value*, char*, G__param*, int) (in /u/apps/root/5.34.13/root/lib/libCint.so.5.34)
==14467==    by 0x6E5A3A0: G__execute_call (in /u/apps/root/5.34.13/root/lib/libCint.so.5.34)
==14467==    by 0x6E5B201: G__call_cppfunc (in /u/apps/root/5.34.13/root/lib/libCint.so.5.34)
==14467==  Address 0xd7636a2 is 0 bytes after a block of size 2 alloc'd
==14467==    at 0x4A0674C: operator new[](unsigned long) (vg_replace_malloc.c:305)
==14467==    by 0x5713757: THcHodoscope::Setup(char const*, char const*) (THcHodoscope.cxx:111)
==14467==    by 0x5713A7E: THcHodoscope::Init(TDatime const&) (THcHodoscope.cxx:142)
==14467==    by 0x4D38C29: THaApparatus::Init(TDatime const&) (THaApparatus.C:193)
==14467==    by 0x4D67E57: THaAnalyzer::InitModules(TList const*, TDatime&, int, char const*) (THaAnalyzer.C:433)
==14467==    by 0x4D68B1F: THaAnalyzer::DoInit(THaRunBase*) (THaAnalyzer.C:667)
==14467==    by 0x4D68126: THaAnalyzer::Init(THaRunBase*) (THaAnalyzer.C:467)
==14467==    by 0x4D6B31A: THaAnalyzer::Process(THaRunBase*) (THaAnalyzer.C:1289)
==14467==    by 0x4E21E97: G__haDict_704_0_6(G__value*, char const*, G__param*, int) (haDict.C:23491)
==14467==    by 0x6DB6148: Cint::G__ExceptionWrapper(int (*)(G__value*, char const*, G__param*, int), G__value*, char*, G__param*, int) (in /u/apps/root/5

THcHodoscope::Clear is called at beginning of analysis before ReadDatabase and Init methods
and therefore  MAXHODHITS and fNPlanes are uninitiated.
Change Clear to ClearEvent
Add call to ClearEvent in Decode
Eliminate Clear method in THcHodoscope.h

    Hardwired numbers are fixed in THcHallcSpectrometer class.
    Generalized functions for center of hodoscope planes and
    their spacings are defined.

Crashfixes0828

Zafar a2 Golden Track

Goldent track is moved to THcHallCSpectrometer::TrackCalc()
  if  fSelUsingScin == 0 then golden track is calculated just
  like podd. i.e. it is the first track with minimum chi2/ndf
  with sorting ON

  if fSelUsingScin == 1 then golden track is calculetd just like
  engine/HTRACKING/h_select_best_track_using_scin.h. This method
  gives the best track with minimum value of chi2/ndf but with
  additional cuts on the tracks. These cuts are on dedx, beta
  and on energy.

  This commit also corrects and modifies the beta code written
  in the THcHodoscope::FineProcess

  At this point we have godlend track and in additoion to
  previous vlues the following values are set to a track:
  Energy, beta, chi2 of beta and dedx[track][first hit].

NOTE: For some tracks we have large differences in value of chi
      square between ENGINE and hcana and this is already
      discussed in the last group meeting. Following emails are
      exchanged regarding this difference of chisquare.

email to Mark:

Hi Mark,
I tried to trace the origin of difference of the chi square value between
ENGINE and hcana. Please see the attached file, it has long table which I
cannot present in this email. Or you can see the file at the following
link, thanks.

https://userweb.jlab.org/~zahmed/chi2diff.txt

Ahmed

Mark's reply:
Ahmed,
        Thanks. I agree with what you see, but there is also another reason
for the difference. For the space point in ch2 the engine and hcana
choose different left/right combinations. That is why you see the large
jump in the summed chi2 from planes 9 and 10. The two codes pick
different left/right because there is a cut on the difference between the
measured stub xp and an expected xp = sp->GetX()/875  and one code has a combination
slightly below the cut value and the other slightly above. I do not see a way
to make the codes agree more exactly since the stub fitting involves a matrix
inversion.

                Cheers,
                       Mark

Goldent track is moved to THcHallCSpectrometer::TrackCalc()
  if  fSelUsingScin == 0 then golden track is calculated just
  like podd. i.e. it is the first track with minimum chi2/ndf
  with sorting ON

  if fSelUsingScin == 1 then golden track is calculetd just like
  engine/HTRACKING/h_select_best_track_using_scin.h. This method
  gives the best track with minimum value of chi2/ndf but with
  additional cuts on the tracks. These cuts are on dedx, beta
  and on energy.

  This commit also corrects and modifies the beta code written
  in the THcHodoscope::FineProcess

  At this point we have godlend track and in additoion to
  previous vlues the following values are set to a track:
  Energy, beta, chi2 of beta and dedx[track][first hit].

NOTE: For some tracks we have large differences in value of chi
      square between ENGINE and hcana and this is already
      discussed in the last group meeting. Following emails are
      exchanged regarding this difference of chisquare.

email to Mark:

Hi Mark,
I tried to trace the origin of difference of the chi square value between
ENGINE and hcana. Please see the attached file, it has long table which I
cannot present in this email. Or you can see the file at the following
link, thanks.

https://userweb.jlab.org/~zahmed/chi2diff.txt

Ahmed

Mark's reply:
Ahmed,
        Thanks. I agree with what you see, but there is also another reason
for the difference. For the space point in ch2 the engine and hcana
choose different left/right combinations. That is why you see the large
jump in the summed chi2 from planes 9 and 10. The two codes pick
different left/right because there is a cut on the difference between the
measured stub xp and an expected xp = sp->GetX()/875  and one code has a combination
slightly below the cut value and the other slightly above. I do not see a way
to make the codes agree more exactly since the stub fitting involves a matrix
inversion.

                Cheers,
                       Mark

Bugs fixed.

  Goldent track is moved to THcHallCSpectrometer::TrackCalc()
  if  fSelUsingScin == 0 then golden track is calculated just
  like podd. i.e. it is the first track with minimum chi2/ndf
  with sorting ON

  if fSelUsingScin == 1 then golden track is calculetd just like
  engine/HTRACKING/h_select_best_track_using_scin.h. This method
  gives the best track with minimum value of chi2/ndf but with
  additional cuts on the tracks. These cuts are on dedx, beta
  and on energy.

  This commit also corrects and modifies the beta code written
  in the THcHodoscope::FineProcess

  At this point we have godlend track and in additoion to
  previous vlues the following values are set to a track:
  Energy, beta, chi2 of beta and dedx[track][first hit].

NOTE: For some tracks we have large differences in value of chi
      square between ENGINE and hcana and this is already
      discussed in the last group meeting. Following emails are
      exchanged regarding this difference of chisquare.

email to Mark:

Hi Mark,
I tried to trace the origin of difference of the chi square value between
ENGINE and hcana. Please see the attached file, it has long table which I
cannot present in this email. Or you can see the file at the following
link, thanks.

https://userweb.jlab.org/~zahmed/chi2diff.txt

Ahmed

Mark's reply:
Ahmed,
        Thanks. I agree with what you see, but there is also another reason
for the difference. For the space point in ch2 the engine and hcana
choose different left/right combinations. That is why you see the large
jump in the summed chi2 from planes 9 and 10. The two codes pick
different left/right because there is a cut on the difference between the
measured stub xp and an expected xp = sp->GetX()/875  and one code has a combination
slightly below the cut value and the other slightly above. I do not see a way
to make the codes agree more exactly since the stub fitting involves a matrix
inversion.

                Cheers,
                       Mark

---------------------------------------------------------------------------------------

This commmit is only for Beta. THcHodoscope is modified and few
bugs are fixed. I will not make pull request for this commit.
Pull request will be made after the completion of Golden track
in THcHallCSpectrometer.

shower counter's FineProcess.

       Remove THcHallCSpectrometer::TrackEnergies method.
       Add GetShEnergy call for each tack in THcShower::fineProcess.

memory consumption with event number.

       Add THcShowerClusterList::purge() method to purge and clean
       THcShower::fClusterList by the end of each event, in
       THcShower::Clear().

crashes.

    Add TrackEnergies method in THcHallCSpectrometer class. It is
    called from TrackCalc metod.

calorimeter.

Add list of clusters fClusterList in THcShower class.
Add method THcShower::GetShEnergy(THaTrack*).
Add method THcShowerClusterList::clear() in THcShowerCluster.h.