Ensure nominal attributes are not lost during alias variable elimination

[Francesco Casella]

As discussed in #4395 and #4510, scaling of variables by means of nominal values can be critical for the performance of nonlinear solvers.
A very common situation is that a trivial equation such as x + y = 0 or x = y is found in a system, where x has a carefully selected nominal attribute, while y doesn't have any.
It would be advisable that in the process of alias variable elimination, variables with nominal attribute are preferably kept, while variables which don't are preferably discarded.

• is this simple logic already implemented in the back-end?
• if not, could you please implement it?

[Patrick Täuber]

The above situation is already handled by the backend.

[Francesco Casella]

Excellent, thanks!