fix: separate scenario from Model object and improve update of soil properties

[nielsvharten-tennet]

Separate scenario from model

Rationale: The model should represent only the static environment and optional persistent configuration. A scenario is execution-specific data and should therefore be supplied per run rather than stored as part of the model state. Keeping scenario as a model attribute complicates state management.

Breaking change: This PR changes the public API of abstract_model and model_factory.

• scenario is now passed as a parameter to model.run() and is no longer stored as an attribute of Model.
• model.set_scenario() has been removed.
• scenario to be passed to model.run() may be unvalidated. Scenario validation is done as first step of model.run().
• temperature_result is no longer stored as attribute of Model; it is now part of the simulation_run.
• Documentation is updated to reflect these changes.

Improve updates of soil properties

• cable._update_capacity_grid() now always updates capacity_grid and no longer checks whether the update is significant. No measurable runtime impact is expected because this operation is not part of finite_difference_matrix_diagonals.
• cable._update_rho_grid() now accepts either a scalar value or a NumPy array of rho values, enabling more efficient updates from cable_soil._update_soil_resistivity().
• cable_soil._update_soil_resistivity() now calls self._update_rho_grid() only once instead of twice. This significantly improves efficiency when soil_drying=True and avoids invalidating finite_difference_matrix_diagonals due to intermediate partial updates.
• Removed model_soil._check_if_daily_update_due(). The method obscured update behavior and provided only limited runtime benefits. Runtime impact is expected only for scenarios where soil_rho fluctuates by more than 1% within a 24-hour period. Users who wish to avoid per-timestep soil_rho updates should preprocess or resample the input scenario accordingly.