wishlist for data weighting options for future lingcod assessments (and SS models in general)

[iantaylor-NOAA]

These are elements left over from #60 which I closed because we've run out of options for this year.
These may be better recorded elsewhere but at least they are noted somewhere for now.

• ability to separate weight discard and retained length comps from the same fleet (discussed in allow separate Dirichlet-multinomial parameter for discard vs retained length comp nmfs-ost/ss3-source-code#93)
• tuning algorithm for discard rates and mean body weight data
• research the impact of altering the input sample sizes (e.g. multiply by 10) in cases where Dirichlet-multinomial weights are estimated near the upper bound leading to 100% of the input sample sizes.