add a test for internetarchive#9808 and internetarchive#9794

unfortunately this test demopnstrates the correct result
of not matching the light record, so it's still unclear why a dated
record matched an undated record with an ISBN.
More investigation is required.

openlibrary/catalog/add_book/tests/test_match.py

@@ -392,7 +392,6 @@ def test_matching_title_author_and_publish_year_but_not_publishers(self) -> None
             'publishers': ['Standard Ebooks'],
             'title': 'Spoon River Anthology',
         }
-
         assert threshold_match(existing_edition, potential_match1, THRESHOLD) is False
 
         potential_match2 = {
@@ -401,6 +400,24 @@ def test_matching_title_author_and_publish_year_but_not_publishers(self) -> None
             'title': 'Spoon River Anthology',
         }
 
-        # If there i s no publisher and nothing else to match, the editions should be
+        # If there is no publisher and nothing else to match, the editions should be
         # indistinguishable, and therefore matches.
         assert threshold_match(existing_edition, potential_match2, THRESHOLD) is True
+
+    def test_noisbn_record_should_not_match_title_only(self):
+        # An existing light title + ISBN only record
+        existing_edition = {
+            # NO author
+            # NO date
+            #'publishers': ['Creative Media Partners, LLC'],
+            'title': 'Just A Title',
+            'isbn_13': ['9780000000002'],
+        }
+        potential_match = {
+            'authors': [{'name': 'Bob Smith'}],
+            'publish_date': '1913',
+            'publishers': ['Early Editions'],
+            'title': 'Just A Title',
+            'source_records': ['marc:somelibrary/some_marc.mrc'],
+        }
+        assert threshold_match(existing_edition, potential_match, THRESHOLD) is False