core: Add more information to the return value of Problem.simulate

[g-braeunlich]

Currently, Problem.simulate() returns a numpy array containing the objective values measuring the performance of the simulated design.
For some problem types it would be beneficial to return additional (problem specific) information.
Goal: Introduce new API / change API so this is possible.

[linear]

SOH-122

[g-braeunlich]

Proposals

In all the following proposals, a new result type is introduced in engibench.core:

@dataclasses.dataclass
class SimulationResult:
    """Base type for the return value of `Problem.simulate_with_details() (or `Problem.simulate()`)`.

    Derived problem classes may return instances of derived `SimulationResult` classes, to return more fields.
    """

    objective_values: npt.NDArray
    """Performance of the simulated design -- each entry corresponds to an objective value."""

Backwards compatible change

We add

    def simulate_with_details(self, design: DesignType, config: dict[str, Any] 
| None = None) -> SimulationResult:
        r"""Launch a simulation on the given design and return the performance.

        Args:
            design (DesignType): The design to simulate.
            config (dict): A dictionary with configuration (e.g., boundary cond
itions, filenames) for the optimization.
            **kwargs: Additional keyword arguments.

        Returns:
            An instance of `SimulationResult` or a derived class.
        """
        return SimulationResult(self.simulate(design, config))

Breaking change (a bit)

We also add Problem.simulate_with_details() but this time:

     def simulate_with_details(self, design: DesignType, config: dict[str, Any] 
| None = None) -> SimulationResult:
        raise NotImplementedError

Then, we also change:

     def simulate(self, design: DesignType, config: dict[str, Any] | None = None) -> npt.NDArray:
-        raise NotImplementedError
+        return self.simulate_with_details(design, config).objective_values

This would require that we add a simulate_with_details() method to all problems.
In contrast to the next proposal, the behavior of Problem.simulate() would not change -> no other changes required.

Breaking change

We change Problem.simulate(), so it returns a SimulationResult instead of a numpy array.
This would require changes in all problems as well as in all places where Problem.simulate() is called.
But we would keep the API simple and at the same time extensible.