`flatten_timepoint_specific_output_overrides` should not do anything if there are no timepoint-specific output parameters

flatten_timepoint_specific_output_overrides's only purpose is getting rid of timepoint-specific output parameters. If there aren't any, it shouldn't modify the petab problem. Currently, the problem is modified nonetheless.

Depending on the exact way of fixing this, unflatten_simulation_df will probably also have to be adapted.

Locate flatten_timepoint_specific_output_overrides and inspect how it changes the PEtab problem when no timepoint-specific output parameters exist. Check unflatten_simulation_df as well, then verify that problems without such parameters remain unchanged while existing flattening behavior still works.