strange/poor fusion and canopy layering with grass and nocomp/fixed-biogeog

[rgknox]

I'm working through tests of the prototype two-stream code. I'm running a 4x5 degree global grid using fixed-biogeography and no-comp. My diagnostics brought me to an unusual patch. There are 20 cohorts all in only 1 canopy layer. All of the cohorts are the same pft (the c4 grass). The first cohort has an LAI of 2.7, and the other 19 have almost the exact same LAI of 2.10 with small deviations past the 3 decimal. They all have the same SAI. The first cohort takes up 97 percent of the canopy area, the rest of them take up the other 3% and nearly add up to 100% of the area: 99.9999999999%.

This is a no-comp run, so it makes sense that the patch is only 1 pft.

I think its odd that there is no second canopy layer, unless we have a special provision where we don't allow a second grass layer (but I can't remember such provisions).

Its also odd that cohorts 2-19 have not been fused, if not all of them. They are almost identical.