Extra parameters to model discards

[gustavdelius]

In addition to the selectivity curve which, when combined with the size distribution of the species, determines the size distribution of the catches, there should also be a selectivity curve that models which of the caught fish are selected to be landed. The size distribution of the landed catch would then be obtained as the product of the size distribution of the catch with this additional landings selectivity curve. I propose that this curve should simply be a sigmoidal curve, parametrised by new gear parameters landings_l50 and landings_l25. These new parameters will be used by getYield() but will be irrelevant for the calculation of the fishing mortality.

In addition to yield_observed we should also have a discard_observed column in the gear_params data frame, giving the rate at which biomass is discarded.

[gustavdelius]

I believe this will require the following:

• In getYieldGear() check if there is a gear parameter landings_l50. If so, set default for landings_l25 if necessary, calculate the landings sigmoid that gives the proportion of the catch that is landed and multiply the current yield by this function before returning.
• Add getDiscardsGear() and getDiscards() functions.
• Add an include_discards = FALSE argument to plotYield(). If called with include_discards = TRUE, add an extra curve for the discards.
• In the documentation for gear_params() mention the new optional landings_l50 and landings_l25 parameters.
• Search for any use of getFMort() that might be used to implement a yield calculation bypassing getYield(). If found, this needs to be changed.
• Add tests
• Update NEWS.md
• Make changes in mizerExperimental

[gustavdelius]

I have decided to develop this in mizerExperimental, see sizespectrum/mizerExperimental#65