Is that just a matter of changing:

self._frozen_resolutions[conf].target_to_resolved_coordinates.get(target)

to:

self._frozen_resolutions[conf].target_to_resolved_coordinates.get(target, [])

Or is there a deeper problem at work here?