test(integ): correct baseline failure docstring, log cleanup failures

Live verification of both integ tests in both directions:
- With fix: 2 passed (38s)
- Baseline v3.16.0: 2 failed as intended. no_s3 test failed at the
  HubAccessConfig assertion (CreateModel accepted the gated gemma-2b
  request without HubAccessConfig; the endpoint then failed
  asynchronously). Aliased test failed at JumpStartConfig validation
  (hub_content_name field does not exist on baseline).

Corrections from the live run: the no_s3 docstring claimed a synchronous
CreateModel ValidationException, but the deterministic failure signal is
the HubAccessConfig assertion. Cleanup except block now logs failures
instead of silently swallowing them (three Failed-status endpoints
leaked during verification runs and were deleted manually).

---

X-AI-Prompt: Run both directions live now
X-AI-Tool: meshclaw

Author: Eli Davidson

sagemaker-serve/tests/integ/test_private_hub_artifact_resolution.py

@@ -389,8 +389,8 @@ def _deploy_and_assert_hub_access_config(
         ):
             try:
                 op(**kwargs)
-            except Exception:
-                pass
+            except Exception as e:
+                logger.warning("Cleanup failed for %s: %s", kwargs, e)
 
 
 @pytest.mark.slow_test
@@ -401,8 +401,10 @@ def test_deploy_with_no_s3_execution_role(
     S3 permissions. Passes only when the SDK attaches HubAccessConfig to
     the CreateModel call (SageMaker brokers artifact access via the hub).
 
-    On v3.16.0 this fails at CreateModel with a ValidationException:
-    'Could not access model data at s3://jumpstart-cache-prod-...'.
+    Verified against the v3.16.0 baseline: fails at the HubAccessConfig
+    assertion (the created Model resource has no
+    S3DataSource.HubAccessConfig, so artifact access is not brokered and
+    the endpoint cannot serve without public-bucket S3 permissions).
     """
     _deploy_and_assert_hub_access_config(
         hub_name=private_hub,