🐛 Crash in optimization_group_calculator_linked when using guidance spectra

[jsnel]

Version information

• pyglotaran version: main

Describe the bug

The code

   dataset["matrix"] = (
                (
                    (self._model_dimension),
                    ("clp_label"),
                ),
                self._group.matrices[label].matrix,
            )

doesn't work for an analysis with multiple dataset, including some guidance spectra (it crashes on creating the result for the first guidance spectra) because self._group.matrices then returns a list of CalculatedMatrix objects, instead of a single CalculatedMatrix object.

A fix would be

 dataset["matrix"] = (
                (
                    (self._global_dimension),
                    (self._model_dimension),
                    ("clp_label"),
                ),
                np.asarray([m.matrix for m in self._group.matrices[label]]),
            )

but only in the case where the result is a list of CalcualtedMatrix objects.

To Reproduce

Run the Artificial Lego example.

Expected behavior

No crashes, while also not failing for all existing examples and unit tests.

Possible fix

Testing what the type is and excuting case dependent code.

 dummy = self._group.matrices[label]
        if isinstance(dummy, CalculatedMatrix):
            dataset["matrix"] = (
                (
                    (self._model_dimension),
                    ("clp_label"),
                ),
                self._group.matrices[label].matrix,
            )
        else:
            dataset["matrix"] = (
                (
                    (self._global_dimension),
                    (self._model_dimension),
                    ("clp_label"),
                ),
                np.asarray([m.matrix for m in self._group.matrices[label]]),
            )