To be called at the start of each analysis.

 To be called at the start of each run.  (Present structure of Pan
 is that each analysis handles only one run, so this is somewhat
 redundant with Init.)  "Prime the pump" by getting a new empty
 pair and making sure fPair is null; set maximum fEDeque and
 fPDeque length, which depends on trigger rate, and maximum
 fEHelDeque length, which depends on delay value and oversampling
 factor.  Find out pair type and set fPairType accordingly.

 To be called at the start second pass of each run.  (Present
 structure of Pan is that each analysis handles only one run, so
 this is somewhat redundant with Init.)  "Prime the pump" by
 getting a new empty pair and making sure fPair is null.

 Main analysis routine -- this is the event loop.  Get an event,
 preprocess it; if the event deque is full, pop off an event,
 analyze it, and pass it to the run to be accumulated; likewise
 for the pair deque.

 To be called at the end of each run.  (Present structure of Pan
 is that each analysis handles only one run, so this is somewhat
 redundant with Finish.)  Clear out the queues, get rid of all
 pairs, and get rid of the pair tree.  But write it out first...

 To be called at the end of each analysis.

 Preprocess event, checking for cut conditions, and putting event
 in helicity delay queue.  If that fills up, take an event off the
 end, insert true helicity, and put it on the event delay
 queue.  Package these events into pairs.  If a pair is completed,
 push it onto pair delay queue.

 Handle analysis of an event that has been through the event delay
 queue, putting results into the event.

 Handle analysis of a pair that has been through the pair delay
 queue, putting results into the pair.

 Look at the pairing type to create a new pair object of the
 correct type

 Routine a derived class can call to do some of the analysis
 automatically.

 Place into the tree space copies, difference, and/or asymmetries
 for channels listed in fTreeList, depending on flags.  Also put
 these into the pair as labelled quantities.

 These lists are created in InitChanLists; different analyses can
 produce different lists, then just call AutoPairAna to handle
 some if not all of the pair analysis.

 At the end, call feedback routines if enabled.

 Last feedback at the end of the run. To have a chance to send
 a feedback, this function needs enough pairs.
 It checks the data vector size and computes an error.

 keep this one aside to the other because the X and Y feedback
 are correlated.
 Send position feedback value to EPICS

 Send position feedback value
 will write the function when hardware defined ....

 Data access functions
 Monitoring related methods