[BUG] ScipyDeSolver (and potentially other solvers) does not work with simulator problems.

[light-weaver]

• Describe the bug
  ScipyDESolver uses PolarsEvaluator to conduct evaluations. This does not handle simulators or surrogates. Replacing with Simulator Evaluator fixed the issue for me.

• To reproduce
  Steps (or a minimal code example) to reproduce the behavior:

# Get metallurgical problem from https://desdeo.readthedocs.io/en/latest/howtoguides/advancedSurrogates/
test = {"UTS":1, "YS":1, "ELON":1} # Just some random values, does not need to be valid
scalarized_problem, symbol = add_asf_nondiff(
        metall_problem, symbol="AAA", reference_point=test, ideal=test, nadir=test, reference_in_aug=True
    )
solver = ScipyDeSolver(scalarized_problem)
#solver.evaluator = SimulatorEvaluator(scalarized_problem)
results: SolverResults = solver.solve(symbol)

Uncommenting the solver.evaluator statement makes the thing work as I need it to. But I do not know if it breaks any other usecases.

• Expected behavior
  Thing should work.

• Actual behavior / error output
  Thing did not work. In PolarsEvaluator._polars_evaluate():

        for symbol, expr in self.objective_expressions:
            if expr is not None:
                # expression given
                ....
            else:
                # expr is None and there are no no simulator or surrogate based objectives,
                # therefore we must get the objective function's value somehow else, usually from data
                ...

The expr variable turns out to be None, which then defaults to using find_closest_points()

• Environment
  • DESDEO version (e.g. output of pip show desdeo): 2.6.0
  • Python version: 3.13.7
  • Operating system: I use arch btw.
  • Installation method (PyPI / from source, uv / poetry / pip): uv (recently synced)
  • Relevant optional solvers installed (Gurobi, IPOPT, BONMIN, CBC, ...): All of them, but this isn't relevant.