Fix AttributeError serializing implicit output collections during job prep

For map-over jobs, `io_dicts(exclude_implicit_outputs=True)` unconditionally
included `output_dataset_collections` (JobToImplicitOutputDatasetCollectionAssociation).
These are shared DatasetCollections with N precreated elements where only the
current job's element has its HDA set. Serializing the full collection during
metadata setup crashes on the N-1 unpopulated elements.

Exclude these implicit output collections when `exclude_implicit_outputs=True`,
consistent with the existing exclusion of output_dataset_collection_instances.

Fixes #22015

Author: mvdbeek

lib/galaxy/model/__init__.py

@@ -1838,7 +1838,8 @@ def io_dicts(self, exclude_implicit_outputs=False) -> IoDicts:
                 if obj.name not in out_data:
                     out_collections[obj.name] = obj.dataset_collection_instance
                 # else this is a mapped over output
-        out_collections.update([(obj.name, obj.dataset_collection) for obj in self.output_dataset_collections])
+        if not exclude_implicit_outputs:
+            out_collections.update([(obj.name, obj.dataset_collection) for obj in self.output_dataset_collections])
         return IoDicts(inp_data, out_data, out_collections)
 
     # TODO: Add accessors for members defined in SQL Alchemy for the Job table and