Develop "harvest version" of multi-population IPM

[ebuhle]

For the US v. OR analysis, I initially considered building a modified version of IPM_adult_pp called IPM_F, customized to the features and questions related to a multi-stock fishery. The idea was to implement realistic, non-fixed rate harvest control rules by making F_rate a function of total abundance in both the retrospective estimation and forward simulation phases. I eventually abandoned this idea when I realized that the method I'd envisioned for estimating F_rate from the time series of observed total catch didn't make sense. The main obstacle was the fact that the set of populations being modeled was not the entirety of the multi-pop aggregate represented in the catch, so you couldn't simply predict the catch by applying F_rate to the summed pre-harvest adult abundance. Likewise, the "independent variable" in the control rule function didn't actually correspond to any state variable in the model.

After expanding the dataset to include Upper Columbia spring Chinook, however, I discovered that the summed pre-harvest adult run size is actually very closely correlated with the TAC time series of aggregate upriver natural spring Chinook, i.e. the "denominator" for estimating harvest rate where the "numerator" is the series of total catch. This opens the possibility of doing the estimation the right way and implementing arbitrary abundance-based control rules in the forward simulation, which would greatly improve the model's realism and generality.

Next steps:

• Double-check the correspondence between summed pre-harvest abundance and the TAC series.
• Estimate F_rate from catch data, compare to TAC-reported estimates.
• Implement harvest control rule in generated quantities block.