Once-through model evaluation doesn't produce inequality constraint residuals

[jonmaddock]

The inequality constraint residuals are all 0.0 in the output mfile when performing once-through (non-optimising) runs.

[timothy-nunn]

I remember fixing a related issue in a path tracker I wrote to see where the solver was walking about, could help (but probably wont because it was a hacky fix to a hacky branch)? b3c60b6

[jonmaddock]

It was actually quite a bit simpler than that, just a silly typo!