Different sizes of collections feed at find_markers make the Galaxy workflow scheduler to fail

[pcm32]

@pinin4fjords has discovered that certain inputs in the cell_type_field for both droplet and smartseq Galaxy workflows bring the workflow scheduler to failure (without much indication in the logs) when this input derives in different sizes of collections being fed as inputs to the find_markers step.

More explicit annotation and a proper default should mitigate this, but the proper thing to happen here is that the Galaxy UI and logs should reflect that it is failing to schedule because the two collections going into the mentioned tool don't match.