Discard mortality median ratio in missing years of discard data (use mean instead)

[jimianelli]

In orig, the discard mortality (by sex) in the directed fishery for the years for which there are no data on discards was based on the annual deviations in fishing mortality multiplied by the median of the ratio of the discard mortality divided by the fully-selected mortality on males in the directed fishery (note that medians are non-differentiable). GMACS failed to include discard for years for which there were no data. This has been corrected and orig modified to work on mean ratios, which should be better behaved computationally.