test: Improve e2e functional tests for classify_single

Better Logging, Enhanced Error Handling, and Updated Expected Outputs

Author: corpo-iwillspeak

examples/classify_single_example.py

@@ -229,7 +229,7 @@ async def main() -> int:
         resize_images=True,
         compress_images=True,
         timeout=30.0,  # Shorter timeout for single requests
-        affiliate="Crisp",
+        affiliate=os.getenv("ATHENA_AFFILIATE", "athena-test"),
         deployment_id="single-example-deployment",  # Not used
     )
 

src/resolver_athena_client/client/transformers/core.py

@@ -50,7 +50,7 @@ def process_image() -> tuple[bytes, bool]:
             # Resize if needed
             if rgb_image.size != _target_size:
                 resized_image = rgb_image.resize(
-                    _target_size, Image.Resampling.LANCZOS
+                    _target_size, Image.Resampling.BILINEAR
                 )
             else:
                 resized_image = rgb_image

tests/functional/e2e/test_classify_single.py

@@ -14,7 +14,7 @@
     load_test_cases,
 )
 
-TEST_CASES = load_test_cases("live_model")
+TEST_CASES = load_test_cases("benign_model")
 
 FP_ERROR_TOLERANCE = 1e-4
 
@@ -59,12 +59,10 @@ async def test_classify_single(
         )
         actual_output = {k: actual_output[k] for k in test_case.expected_output}
 
-        max_diff = max(
-            abs(test_case.expected_output[label] - actual_output[label])
-            for label in test_case.expected_output
-        )
-        assert max_diff < FP_ERROR_TOLERANCE, (
-            "Output weights differ from expected by more than",
-            f" {FP_ERROR_TOLERANCE}: ",
-            f"expected={test_case.expected_output}, actual={actual_output}",
-        )
+        for label in test_case.expected_output:
+            expected = test_case.expected_output[label]
+            actual = actual_output[label]
+            assert abs(expected - actual) < FP_ERROR_TOLERANCE, (
+            f"Weight for label '{label}' differs by more than {FP_ERROR_TOLERANCE}: "
+            f"expected={expected}, actual={actual}, diff={abs(expected - actual)}"
+            )

tests/functional/e2e/testcases/benign_model/expected_outputs.json

@@ -1,11 +1,11 @@
 {
     "classification_labels": [
-        "Lakeside",
-        "Mountains",
-        "Trees",
-        "Seashore",
-        "Ducks",
-        "Goose"
+        "UnknownCSAM-Lakeside",
+        "UnknownCSAM-Mountains",
+        "UnknownCSAM-Trees",
+        "UnknownCSAM-Seashore",
+        "UnknownCSAM-Ducks",
+        "UnknownCSAM-Goose"
     ],
     "images": [
         [