Fixes None tracing outputs attempted to be concatenated

[MaxGamill-Sheffield]

Apologies, the previous PR was a tad early, I thought the "merge when ready" would wait for until tests finished...

This should provide a hotfix to the errors and similar like them:

disordered_trace_results[str(img)] = disordered_trace_result.dropna(axis=1, how="all")
                                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AttributeError: 'NoneType' object has no attribute 'dropna'

This occurred because the disabled modules would return None instead of an empty dataframe and we expected to have the dataframe function DataFrame.dropna() which obvs didn't exist.

This fix should ensure that dataframes are returned by disabled modules.

It also adds another exception I found when forcing the disordered traces step to fail causing KeyErrors in NodeStats and Ordered Tracing as they couldn't find the image name keys of a skeleton to grab.

[SylviaWhittle]

Very much approve of the different column_sets. This makes things really clear IMO. Also love the returning an empty dataframe that is created in the same return statement