TrajectoryData: Fix bug in get_step_data

[sphuber]

Fixes #4170

The get_step_data would raise a NameError in case a TrajectoryData does not define the cells array. The variable cell would be returned but it would only be defined if self.get_cells() does not return None.

[ltalirz]

haven't looked at the code in this PR, but possibly related: https://github.com/aiidateam/aiida-core/pull/5394/files

I was waiting for some data input there; perhaps we can get that one across the finish line as well with some sample data

[sphuber]

I was waiting for some data input there; perhaps we can get that one across the finish line as well with some sample data

Looking at the linked issue it doesn't seem to be related. This issue solves a bug with the TrajectoryData class itself, whereas the issue of your PR has to do with the various external programs that can be called through the verdi data trajectory show command.

Don't think that you actually need real world data for the tests right? The issue is that exceptions are raised, not that the data is plotted incorrectly. Think you can just mock a trajectory with a fixture.

[giovannipizzi]

Thanks, the implemented behaviour seems OK to me!