imposex assessments (ordinal data) - all measurements are zeros or max value

[RobFryer]

Routines cstm.VDS.index.opt and ctsm.VDS.index.cl are used for ordinal data (imposex assessments) for estimating the cut points when transitioning from one class to another and for putting confidence intervals around the VDSI estimates for each station, year, species combination. Currently they have ad-hoc methods to dealing with the special cases when either:

• all the individual VDS measurements for a station, year, species combination are zero (so the VSDI = 0)
• all the individual VDS measurements for a station, year, species combination are on their upper bound (so the VDSI = 6 for e.g. Nucella)

This has several potential consequences:

• makes convergence more complicated
• puts incorrect confidence intervals around the VDSI estimates in the special cases; in particular the upper confidence limit (when the VDSI is zero) is likely to be too high