Some minor cleanups following #308

[spencerkclark]

In porting the changes in #308 to SHiELD, I noticed a few areas where things could be cleaned up:

• Within the code to coarse-grain tracer fields on surfaces of constant pressure, we can remove the need to look up the tracer name within the prognostic tracer loop since we no longer need to distinguish between the weights for the cloud fraction and hydrometeor tracers.
• Within the code to coarse-grain diagnostics we can make the subroutines for determining whether we need to allocate memory for masked area or mass arrays more readable.

These changes do not change the behavior or answers in any way.