Add new ntuple variables
goodAdcMult == the number of hits in raw ADc for that channel
goodAdcHitUsed == which hit number of the raw ADC hits passed the
                      the ADC time window cut

Modified fAdcTimeWindowMin  and fAdcTimeWindowMax to be arrays
  that are the size of the number of Cerenkov pmts.

  So that the counts can be reported in report files.

   First hit above this cut is taken as the reference time.
   Cuts are per detector and read as an optional parameter in
   each detector class.  TDCs and Flash ADC times have separate cuts.

Implement a second method of selecting best reference time.
  If InitHitList is passed a possitive reference time cut,
    use the first reference time above that cut.  If none of the
    hits pass the cut, then no reference time is set
  If InitHitList is passed a negative cut, the cut is taken to be
    the absolute value of the cut, but a reference time is garuanteed
    to be set as long as there is at least one reference time hit.
    If none of the reference time hits pass the cut, then the last
    hit is taken to be the reference time.  (Should be the largest.)

Allow reference time cuts for all the spectrometer detector classes
  Hodoscopes, drift chambers, Aerogels, Gas Cherenkovs, Calorimeters

  Add End method to each detector that calls MissReport which
  prints the number of events with missing TDC and ADC reference times.

Add variable for the difference between the Hodoscope start time
and the ADC pulse time.

Select good adc hit based on window on this difference

  Each detector object will check the global variable X.present each
  event.  If X.present is false, it indicates that the parent
  spectrometer object determineted that this spectrometer is not in
  this event, and therefor reference times are not expected.  If
  X.present is false, then the reference time warning messages will
  be suppressed.

Moved filling of  fNRegions from parameter to before the fRegionValue array
is defined. The code had been using default fNRegions=4 regardless
of vaule  of  h_num_regions in parameter file.

Modified so that the fGoodPosAdcPulseTime, fGoodNegAdcPulseTime
  or fGoodAdcPulseTime is set to kBig

Modified THcShowerRawHit::SetReference to takeADC signals

Modified THcTrigRawHit::SetReference to take ADC signals

Modified THcShowerArray::ProcessHits so that fAdcTdcOffset add to frAdcPulseTime

Modified THcShowerArray::ReadDatabase so that defaults for fAdcThreshold etc
  are before gHcParms->LoadParmValues

Fix typo in previous commit's THcCherenkov.cxx

Include methods for FADC conversions and apply to non-raw data objects in THcRawAdcHit

Add raw pulse time data object to Cherenkov detector classes

Add raw pulse time data object to shower detector classes

Add raw pulse time data object to hodoscope detector class

   Gets rid of vars.push_back(RVarDef ... stuff

Also do some whitespace cleanup

New parameters are added.  hallc_replay of 3/31/2017 or later is needed.

As of now, the size of the TClonesArray objects are hard-coded in the
respective header files. This is not okay in the long run. Ideally,
the size of the TClonesArrays should be defined by the parameter which
dictates how many readout detectors exist for the respective
detector. I tried doing this in the Init method however, podd
complained regarding the size of the TClonesArrays in the DEF-file
histograms. I was unsuccessful in my first attempt. This should be
addressed in the future even though things work for now.

Add array fADC_hit that is a flag if there is a hit in the ADC
fADC_hit is set in Decode method
The FADC has a threshold on accepting ADC hit
Modified CoarseProcess method to eliminate old FASTBUS bias in the analysis

The calls to `Find()` need to be wrapped in checks if returned pointer is null.
Otherwise a further call to get value produces SegFault. The `LoadParmValues()`
does not suffer from this, since it throws error if non-optional parameter is
not found. The error message also tells which parameter is missing.

I still left three calls to `Find()` in `THcAnalyzer.cxx`. They are wrapped in
checks and some other things...

 - Added new leafs to the hodoscope tree.
 - Added new leafs to the aerogel tree.
 - Added new leafs to cerenkov tree.
 - Added new leafs to the shower tree.

Left all the old leafs in there. Will need a careful cleanup...

This makes the code nicer and with less duplication in logic.

Along with that I had to correct all the call sites to the RawHodoHit in all the
classes that use it.

  Don't check if UInt_t's are >=0
  Remove unused/repeated variables
  Cast enums to Int_t in switch statements
  Initialize temp variables

  Otherwise InitHistlist doesn't have the detector map
  and can't do setup of reference time stuff or sanity checks on the
  detector mapping

Some detector classes lacked the ability to get the appropriate
Apparatus character and were only working for HMS.

  May want to use a hit class that doesn't inherit from THcHodoscope
  as the Cerenkov is not using Neg tubes or TDCs.

  Modify THcFormula and THcParmList to match changes to THaFormula
     (needed for reports)
  Small changes in THcInterface and detector classes for compatibility
     with new OO decoder and podd 1.6

  Don't start variables that are not member variables with f
  Declare temporary variables as locally as possible
  Remove some unused member variables from THcHodoscope.h

      Declare loop variables in the for statement
      Go back to UInt_t for some variables
      Don't prefix non-member variables with f

  Adds tests to THcHodoscope fiducial tracking efficiency
  Adds tests to THcCherenkov for Cerenkov efficiency

  For the most part this is making sure that the loop variable in
  a for loop is of the same type as the variable that it is compared
  against (Often a UInt_t).

This reverts commit e9ce1040.

Conflicts:
	src/THcHallCSpectrometer.cxx