Changes (bloss, Apr 09):

 + Modified scheme to use total water (vapor + cloud liquid) as a
prognostic variable in place of separate vapor and cloud liquid
variables.  This should reduce the cost associated with scalar
advection.  With this scheme, saturation adjustment occurs before
calling the microphysics, so that a consistent set of temperature,
water vapor and cloud liquid values can be input to the microphysics.
This saturation adjustment could be more expensive than the old one
within the microphysics, since you do not have a reference value of
cloud liquid to start from.

 + The latent heating due to the microphysics is not readily
accessible with the total water formulation, so that it is no longer
output in the statistics.

 + The EFFRQ* effective radius outputs in the past were susceptible to
errors due to non-uniformity of hydrometeor fractions across
processors or time steps.  As a result, these outputs have been
eliminated.  In their place, a new output Q*OEFFR which represents a
scaled version of the layer optical depth has been added.  This can be
used to compute a consistent effective radius from the averaged
optical depth and mixing ratio.  For example, the cloud liquid
effective radius may be computed as:

  EFFRC = QC / QCOEFFR

where QC is in g/kg, QCOEFFR is in g/kg/micron and EFFRC is in
microns.

 + The default setting for dopredictNc was changed to .false.
Now, by default, the cloud liquid droplet concentration will be set to
Nc0 which itself has a default value of 100./cm3.
