HierarchicalAmiciCalculator.__call__ missing 'hess' in returned dict upon simulation failure

[dweindl]

I finally got to the bottom of #1147

The problem is that in case of simulation failures HierarchicalAmiciCalculator.__call__ does not set inner_result[HESS] here, although the Hessian was requested:

pyPESTO/pypesto/hierarchical/calculator.py

Lines 113 to 122 in 883138a

	# if any amici simulation failed, it's unlikely we can compute
	# meaningful inner parameters, so we better just fail early.
	if any(rdata.status != amici.AMICI_SUCCESS for rdata in inner_rdatas):
	# if the gradient was requested, we need to provide some value
	# for it
	if 1 in sensi_orders:
	inner_result[GRAD] = np.full(
	shape=len(x_ids), fill_value=np.nan
	)
	return inner_result

This results in the following error:

  File "pypesto/optimize/task.py", line 59, in execute
    optimizer_result = self.optimizer.minimize(
                       ^^^^^^^^^^^^^^^^^^^^^^^^
  File "pypesto/optimize/optimizer.py", line 167, in wrapped_minimize
    result = minimize(
             ^^^^^^^^^
  File "pypesto/optimize/optimizer.py", line 136, in wrapped_minimize
    result = minimize(
             ^^^^^^^^^
  File "pypesto/optimize/optimizer.py", line 83, in wrapped_minimize
    result = minimize(
             ^^^^^^^^^
  File "pypesto/optimize/optimizer.py", line 1342, in minimize
    opt.minimize(x0)
  File "fides/minimize.py", line 333, in minimize
    funout_new = self.fevaler(x_new)
                 ^^^^^^^^^^^^^^^^^^^
  File "fides/minimize.py", line 40, in __call__
    ret = self.fun(x, **self.funargs)
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "pypesto/objective/amici/amici.py", line 395, in __call__
    return super().__call__(x, sensi_orders, mode, return_dict, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "pypesto/objective/base.py", line 197, in __call__
    result = ObjectiveBase.output_to_tuple(
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "pypesto/objective/base.py", line 334, in output_to_tuple
    output += (kwargs[HESS],)
               ~~~~~~^^^^^^
KeyError: 'hess'