dimension of arrays for function calls

[wildsm]

Given n decision variables, the intent is for ibcdfo solvers to pass a single point as an n-by-1 array (rather than 1-by-n).

The distinction may always not be needed, but is intended to allow poptus functions that are called to be vectorized in the sense that when a p-by-n input is passed (for p>1), this means that there are p outputs of the (vector- or scalar-valued) function