Fix set_peek method for copied datasets

dataset_assoc is the job to ouptut dataset association that references
the HDA created by a job. If the HDA is copied the copies need to be
updated as well, so here we need to call `set_peek` on dataset, not
dataset_assoc.dataset.

Broken in f4343d876f

Thanks for tracking this down @jmchilton!

lib/galaxy/jobs/__init__.py

@@ -1399,10 +1399,10 @@ class JobWrapper(HasResourceParameters):
                     line_count = context.get('line_count', None)
                     try:
                         # Certain datatype's set_peek methods contain a line_count argument
-                        dataset_assoc.dataset.set_peek(line_count=line_count)
+                        dataset.set_peek(line_count=line_count)
                     except TypeError:
                         # ... and others don't
-                        dataset_assoc.dataset.set_peek()
+                        dataset.set_peek()
                 else:
                     # Handle purged datasets.
                     dataset.blurb = "empty"