FATES simulations need some help with the processor layout "decomposition"

### Brief summary of bug

There was a discussion about FATES performance. With notes here..

https://docs.google.com/document/d/112_UUcmfo8Li1Avw3b8bAz5nAgAioR5h08cBRO6H-B

There's a lot of work to look into how to improve photosynthesis for FATES. But, something else that came
out here is that CTSM is poorly load balanced for FATES. From slides that @rgknox showed time running
photosynthesis for a FATES case varied from 70 seconds to 3000. Which indicates that the load balance
is really poor.

### General bug information

**CTSM version you are using:** Probably all versions

**Does this bug cause significantly incorrect results in the model's science?** No

**Configurations affected:** FATES

### Details of bug

We set the decomposition at initialization. For BGC cases our current decomposition seems to perform fine. Part of
what it does is to try to get land points on opposite sides of the globe which handles day/night issues. It's likely that
FATES needs to also have something that gives weighting toward the complexity of gridcells. So possibly based on
the number of cohorts and patches within a gridcell. This could be done in a static way at initialization by having an
average "map" of this for simulations that are spunup. Another way and one that might be required by FATES is to
do the load balancing on the fly as the simulation moves forward.