Switch V2 pagination back to using IDs, which should be much faster and easier on the DB

Also adds a test for the tags endpoint

test/registry/registry_tests.py

@@ -16,6 +16,7 @@ from test.registry.protocol_fixtures import *
 from test.registry.protocols import Failures, Image, layer_bytes_for_contents, ProtocolOptions
 
 from app import instance_keys
+from data.model.tag import list_repository_tags
 from util.security.registry_jwt import decode_bearer_header
 from util.timedeltastring import convert_to_timedelta
 
@@ -718,6 +719,31 @@ def test_catalog(public_catalog, credentials, expected_repos, page_size, v2_prot
     assert set(expected_repos).issubset(set(results))
 
 
+@pytest.mark.parametrize('username, namespace, repository', [
+  ('devtable', 'devtable', 'simple'),
+  ('devtable', 'devtable', 'gargantuan'),
+  ('public', 'public', 'publicrepo'),
+  ('devtable', 'buynlarge', 'orgrepo'),
+])
+@pytest.mark.parametrize('page_size', [
+  1,
+  2,
+  10,
+  50,
+  100,
+])
+def test_tags(username, namespace, repository, page_size, v2_protocol, liveserver_session,
+              app_reloader, liveserver, registry_server_executor):
+  """ Test: Retrieving results from the V2 catalog. """
+  credentials = (username, 'password')
+  results = v2_protocol.tags(liveserver_session, page_size=page_size, credentials=credentials,
+                             namespace=namespace, repo_name=repository)
+
+  expected_tags = [tag.name for tag in list_repository_tags(namespace, repository)]
+  assert len(results) == len(expected_tags)
+  assert set([r for r in results]) == set(expected_tags)
+
+
 def test_pull_torrent(pusher, basic_images, liveserver_session, liveserver,
                       registry_server_executor, app_reloader):
   """ Test: Retrieve a torrent for pulling the image via the Quay CLI. """