Make CLM-FATES run faster in NorESM

[rosiealice]

The model runs slowly (metrics to be formalized as part of this issue).

Ideas for speeding it up include

1. Reducing the number of patches by altering the tolerance thresholds
2. Rmoving the new FATES shrub PFT at high latitudes.
3. Modifying the tolerances on photosynthetic iteration loops.

To test any of these, we need some kind of standard measure of performance. FATES changes performance based on the compleity of the ecosystem, and so the equilibrium speed is not known for some time into the simulation.
But the speed of the model in the early stages is also relevant. I will look into how to determine the model speed as part of this discussion.