The model stops because of a new fatal warning that was added to
detect erroneous situations with multiple real observations in a superperiod.
With the retroyr, there are zero real obs in the terminal
super-years. However the warning condition was set to <>1,
rather than >1, so the model stops.
So, this should be a rather easy fix in the preliminary calcs section
of the code.
We are in the midst of testing the pre-release for 3.30.19. We
should be able to get this fix into the full release in a couple of
weeks and can make it available in the github action sooner.
Rick