inconsistency between dlower_vai indexing and arrays that use it

[rgknox]

Various arrays we have are dimensioned by leaf layer. Leaf layer boundaries are defined by top-down depth integrated vegetation area, and controlled by the input parameters that define the starting bin width and an increase factor.

These parameters define the array dlower_vai, set here:
https://github.com/NGEET/fates/blob/main/main/FatesInterfaceMod.F90#L928-L937

The issue is that when we determine the number of leaf layers there are in a cohort, if the cohort has more VAI than the last bin in dlower_vai, it will generate an index that is out of bounds for various arrays. See here:

https://github.com/NGEET/fates/blob/main/biogeochem/EDCanopyStructureMod.F90#L2258

This is because dlower_vai is the size nlevleaf and so are the the arrays that use dlower_vai as its indexing... such as ts_net_uptake, see here:

https://github.com/NGEET/fates/blob/main/main/EDParamsMod.F90#L113
https://github.com/NGEET/fates/blob/main/biogeochem/FatesCohortMod.F90#L166

I think there are at least two ways to improve this. We could increase the allocation of arrays that use this indexing to nlevleaf+1, or we could force the last index of dlower_vai to be very very large (like e+20) or something silly large.