[Halld-offline] Recent BCAL software changes

[Will Levine]

A while back I reported on an issue in the mcsmear BCAL simulation
that degraded the timing/z-resolution and clustering performance. The
full details are here:
https://halldweb1.jlab.org/wiki/index.php/BCAL_reconstruction_01/16/2012

Last week, I checked in a temporary fix to this problem (r9233): if a
noisy SiPM and a SiPM with real signal are being summed in the same
readout channel, sum the energies, but only consider the timing of the
real hit. David and I agreed that that this would be an OK temporary
fix until we have a more realistic simulation.

The b1pi tests show that this caused a ~6% reduction in total
reconstructed photons. (
https://halldweb1.jlab.org/b1pi/2012-06-18/Linux_RHEL5-i686-gcc4.1.2/
vs. https://halldweb1.jlab.org/b1pi/2012-06-21/Linux_RHEL5-i686-gcc4.1.2/
)

Will