Resolve LDDA <-> Library Dataset circular dependency

This eliminates a lot of excessive flushes that were needed to persist
new `LibraryDatasetDatasetAssociation` (LDDA) and `LibraryDataset` (LD) objects.  These
intermittent flushes would make LD objects visible before
they are associated with a LDDA, causing the
Exception in
https://github.com/galaxyproject/galaxy/pull/10556#issuecomment-717175641

```
galaxy.web.framework.decorators ERROR 2020-10-27 17:26:25,817 Uncaught exception in exposed API method:
Traceback (most recent call last):
  File "/Users/mvandenb/src/galaxy/lib/galaxy/web/framework/decorators.py", line 294, in decorator
    rval = func(self, trans, *args, **kwargs)
  File "/Users/mvandenb/src/galaxy/lib/galaxy/webapps/galaxy/api/folder_contents.py", line 111, in index
    dataset = content_item.library_dataset_dataset_association.dataset
AttributeError: 'NoneType' object has no attribute 'dataset'
```

The trick here is to use `post_update=True`, so that ldda and ld rows
are INSERTED, followed by an UPDATE to create the link.
This is explained at
https://docs.sqlalchemy.org/en/13/orm/relationship_persistence.html.

This is also the groundwork for speeding up library additions by
batching flushes.

test/unit/data/test_galaxy_mapping.py

@@ -1,4 +1,5 @@
 # -*- coding: utf-8 -*-
+import collections
 import unittest
 import uuid
 
@@ -154,6 +155,36 @@ class MappingTests(unittest.TestCase):
         assert isinstance(history.get_display_name(), text_type)
         assert history.get_display_name() == u'Hello₩◎ґʟⅾ'
 
+    def test_hda_to_library_dataset_dataset_association(self):
+        u = self.model.User(email="mary@example.com", password="password")
+        hda = self.model.HistoryDatasetAssociation(name='hda_name')
+        self.persist(hda)
+        trans = collections.namedtuple('trans', 'user')
+        target_folder = self.model.LibraryFolder(name='library_folder')
+        ldda = hda.to_library_dataset_dataset_association(
+            trans=trans(user=u),
+            target_folder=target_folder,
+        )
+        assert target_folder.item_count == 1
+        assert ldda.id
+        assert ldda.library_dataset.id
+        assert ldda.library_dataset_id
+        assert ldda.library_dataset.library_dataset_dataset_association
+        assert ldda.library_dataset.library_dataset_dataset_association_id
+        library_dataset_id = ldda.library_dataset_id
+        replace_dataset = ldda.library_dataset
+        new_ldda = hda.to_library_dataset_dataset_association(
+            trans=trans(user=u),
+            target_folder=target_folder,
+            replace_dataset=replace_dataset
+        )
+        assert new_ldda.id != ldda.id
+        assert new_ldda.library_dataset_id == library_dataset_id
+        assert new_ldda.library_dataset.library_dataset_dataset_association_id == new_ldda.id
+        assert len(new_ldda.library_dataset.expired_datasets) == 1
+        assert new_ldda.library_dataset.expired_datasets[0] == ldda
+        assert target_folder.item_count == 1
+
     def test_tags(self):
         model = self.model