logistic function for division between metabolic and structural litter pools #1010
Description
At the moment, a linear function is used to divide input plant biomass between the metabolic and structural litter pools. This means that for some lignin/nitrogen/phosphorus values a negative amount of plant input can be given to the metabolic litter pool (which the litter model raises an error in response to).
Generally, the above shouldn't happen for plausible lignin/nitrogen/phosphorus values, so I was initially minded to ignore this. However, in the litter model parameterisation @hrlai has found that fitting a logit function helps with stopping the parameterisation from getting stuck on the prior. (which I guess makes sense as under the old model a lot more parameter combinations that produce non-sensical results were possible).
For this reason, we will have to switch using a logistic/sigmodial function for the division between pools (similar to the function used for CUE #746)