level 1 warning as first step toward more in-depth search

https://github.com/OpenMx/OpenMx/blob/master/R/MxModelFunctions.R#L127-L137
Add test script checking returning first row behavior?
What is proper behavior for this?
What is the value of submodel.likelihood + submodel. A[1,1] when submodel.A is “data.X”?
It depends on how many rows in are data. If nrow(data) == 1 then data.X is the data in the first row. If nrow(data) > 1 then data.X is set to NaN. (initial compute)
NaN is dangerous because OpenMx matrix algebras are NA/NaN unsafe.
Throw an error Do this. [Joshua will attempt] → difficult; how about switching to safe matrix multiply when definitions variables are used? slower but safe. we can add an “unsafe definition variables” option?
Just use first row of data anyway, but throw warning. No, problematic
Return vector? Or vector of matrices
After backend processing, algebras are recomputed with all def vars set to the first row.