test(gooddata-eval): assert the verdict per malformed ranking filter

`assert ok == (len(errors) == 0)` was a tautology against an implementation that
returns exactly `len(errors) == 0, errors`, so the case list could not fail:
`attribute=[]` or `using=None` being accepted with ok=True would have gone
unnoticed. Each case now carries its expected verdict.

Confirmed non-vacuous by mutation: making a non-string `attribute` skip instead
of erroring now fails the test, where before it passed.

JIRA: QA-28615
risk: nonprod

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

Author: myhoai

packages/gooddata-eval/tests/test_scoring.py

@@ -132,19 +132,25 @@ def test_ranking_attribute_omitted_does_not_mask_a_wrong_dimension():
 
 
 def test_validate_cross_references_never_raises_on_empty_or_none_uris():
-    """Each of these used to raise AttributeError/TypeError instead of returning a score."""
+    """Each of these used to raise AttributeError/TypeError instead of returning a score.
+
+    Every case carries its expected verdict: `attribute` is optional so None/"" are valid,
+    while a non-string attribute or a missing/None `using` must be reported as an error.
+    Asserting the verdict is what stops a malformed filter from silently passing as valid.
+    """
     cases = [
-        {"type": "ranking_filter", "using": "m_sales", "top": 5, "attribute": None},
-        {"type": "ranking_filter", "using": "m_sales", "top": 5, "attribute": ""},
-        {"type": "ranking_filter", "using": "m_sales", "top": 5, "attribute": []},
-        {"type": "ranking_filter", "using": None, "top": 5},
-        {"type": "ranking_filter", "top": 5},
+        ({"type": "ranking_filter", "using": "m_sales", "top": 5, "attribute": None}, True),
+        ({"type": "ranking_filter", "using": "m_sales", "top": 5, "attribute": ""}, True),
+        ({"type": "ranking_filter", "using": "m_sales", "top": 5, "attribute": []}, False),
+        ({"type": "ranking_filter", "using": None, "top": 5}, False),
+        ({"type": "ranking_filter", "top": 5}, False),
     ]
-    for rank in cases:
+    for rank, expected_ok in cases:
         viz = _viz(query={"fields": _M, "filter_by": {"f_rank": rank}})
         ok, errors = validate_cross_references(viz)
         assert isinstance(ok, bool) and isinstance(errors, list), rank
-        assert ok == (len(errors) == 0), rank
+        assert ok is expected_ok, rank
+        assert bool(errors) is not expected_ok, rank
 
 
 def test_validate_cross_references_accepts_omitted_attribute_but_flags_missing_using():