Hi,
Right, when you export a source's labelset to CSV, that CSV data references the ID numbers of the CoralNet labels (which are defined globally, not on the source level). So when you try to import from that labelset CSV-file, it expects those same labels to exist in the CoralNet database with the same IDs.
Note that in general, we don't have the resources to support local builds of CoralNet, so in situations like this you'll need some extra effort to make things work for you. I'd suggest going to the production label list page (
https://coralnet.ucsd.edu/label/list/ ) and scraping the label names and functional groups you're interested in. Then write some lines of Python code that you can run in manage.py shell to create the functional groups, followed by the labels.
Hope that at least helps a bit.