test: close coverage gaps and add frontend test infrastructure

Backend (516 tests, 94% coverage):
- Add azure_openai endpoint/deployment validation tests (config.py -> 100%)
- Add _total_conversations and _avg_turns direct tests (queries.py -> 100%)
- Add transformer edge cases: list content, string checkpoint, invalid JSON,
  malformed message graceful skip (transformer.py -> 93%)
- Add safety combined status_code+error_message interaction tests
- Fix ambiguous 200/422 assertion to strict 422
- Add E2E pagination shape assertions (total, page, per_page, row count)
- Fix ReplayPool mock to respect LIMIT/OFFSET params

Frontend (23 tests, vitest + happy-dom + @testing-library/react):
- Add vitest infrastructure with happy-dom environment
- Add api.ts tests: success, HTTP error, success=false, URL encoding
- Add DashboardPage tests: loading, data, error, empty states
- Add ReplayListPage tests: loading, empty, data, error, status badge classes
- Add ReplayPage tests: loading, steps, empty, error states

backend/tests/unit/replay/test_transformer.py

@@ -153,3 +153,105 @@ class TestTransformCheckpoints:
         rows = [_make_row([{"type": "human", "content": "Hi"}])]
         steps = transform_checkpoints(rows)
         assert isinstance(steps[0].timestamp, str)
+
+    def test_list_content_joined_to_string(self) -> None:
+        from app.replay.transformer import transform_checkpoints
+
+        rows = [
+            _make_row(
+                [
+                    {
+                        "type": "human",
+                        "content": [
+                            {"text": "Hello"},
+                            {"text": " world"},
+                        ],
+                    }
+                ]
+            )
+        ]
+        steps = transform_checkpoints(rows)
+        assert len(steps) == 1
+        assert steps[0].content == "Hello  world"
+
+    def test_checkpoint_as_string_skipped(self) -> None:
+        from app.replay.transformer import transform_checkpoints
+
+        rows = [
+            {
+                "thread_id": "t1",
+                "checkpoint_id": "cp1",
+                "checkpoint": "not-a-dict",
+                "metadata": {},
+            }
+        ]
+        steps = transform_checkpoints(rows)
+        assert steps == []
+
+    def test_channel_values_not_dict_skipped(self) -> None:
+        from app.replay.transformer import transform_checkpoints
+
+        rows = [
+            {
+                "thread_id": "t1",
+                "checkpoint_id": "cp1",
+                "checkpoint": {"channel_values": "bad"},
+                "metadata": {},
+            }
+        ]
+        steps = transform_checkpoints(rows)
+        assert steps == []
+
+    def test_tool_result_valid_json_parsed(self) -> None:
+        from app.replay.transformer import transform_checkpoints
+
+        rows = [
+            _make_row(
+                [
+                    {
+                        "type": "tool",
+                        "content": '{"order_id": "123", "status": "shipped"}',
+                        "name": "get_order_status",
+                    }
+                ]
+            )
+        ]
+        steps = transform_checkpoints(rows)
+        assert len(steps) == 1
+        assert steps[0].result == {"order_id": "123", "status": "shipped"}
+
+    def test_tool_result_invalid_json_wrapped(self) -> None:
+        from app.replay.transformer import transform_checkpoints
+
+        rows = [
+            _make_row(
+                [
+                    {
+                        "type": "tool",
+                        "content": "not valid json",
+                        "name": "some_tool",
+                    }
+                ]
+            )
+        ]
+        steps = transform_checkpoints(rows)
+        assert len(steps) == 1
+        assert steps[0].result == {"raw": "not valid json"}
+
+    def test_malformed_message_skipped_gracefully(self) -> None:
+        from app.replay.transformer import transform_checkpoints
+
+        rows = [
+            _make_row(
+                [
+                    {"type": "human", "content": "Good message"},
+                    42,  # not a dict -- will raise in _step_from_message
+                    {"type": "ai", "content": "Response", "tool_calls": []},
+                ]
+            )
+        ]
+        steps = transform_checkpoints(rows)
+        # The malformed message is skipped; the other two produce steps.
+        assert len(steps) == 2
+        assert steps[0].step == 1
+        assert steps[1].step == 2