PlantsModel - moving to clear cohort / cell calculations

The current implementation nests cohort data within communities, but actually most calculations can be done in parallel across all cohorts, with cell_id recorded in self.cohorts. This should give clearer and faster code by restricting the amount of time spent in cell_id loops.

• The communities attribute disappears.
• The following attributes go from dict[int, XYZ] to just XYZ
  • self.biomasses (This change probably has the most engineering and, since we want to rework the internals of Biomasses anyay, probably best done as a separate issue, although the temporary glue holding the join together would be finicky)
  • self.stem_allocations
  • self.growth_increments
  • self.per_stem_gpp (probably becomes a field in self.cohorts)
  • self.per_stem_transpiration (also probably becomes a field in self.cohorts)
• The self.canopies attribute stays as dict[cell_id, Canopy]

The update flow would be:

1. Mortality - applied across all cohorts (drop current cell_id loop)
2. Recruitment - applied across all cohorts (drop current cell_id loop)
3. Update allometry - a single StemAllometry object across cohorts. (drop current cell_id loop updating community objects).
4. Herbivory - applied across all cohorts
5. Calculate canopies - CELL level calculations (store cohort indices on canopy object).
6. Update canopy layers - CELL level calculations.
7. Subcanopy light capture - already in parallel across cells
8. Calculate light use efficiency (already at cell by layer level).
9. Estimate GPP - CELL level calculations, extract cohorts in each cell,
10. Calculate nutrients - applied across all cohorts
11. Allocate GPP - applied across all cohorts
12. Subcanopy dynamics - already in parallel