General
-------

A weird behavior when submitting batch jobs. They seem to get
corrupted if more than one is running. Why?


Parallel 
--------

 - 410: Prior to 410 the NN corrected gamma energy and positions were assumed
   to not change much with the calibration coefficients. This was a bad idea.
   The NN corrections are now recalculated along with the moments in
   Pi0Calibration::CorrectClusterBlocks().


 - 412: Noticed an oscillation around the mass. Maybe we should take smaller
   epsilon steps? Epsilon fraction was set 0.02. 

 - 413: Epsion fraction now set to 0.01 

 - 412 Comparison: Something is way wrong. Either there are lots of
        bad events (due to noisy channels or loose trigger), or something is
        wrong in the code.

 - Somewhere around here the data was replayed and calibrations got mixed up!
 
 - Going to replay all the parallel data using series 44. 

 - Done replaying with series 44. Going to us calibration compare #420 as my
   base line for comparing subsequent calibrations. 

 -