fix(evidence): resolve final review findings

harness/tests/test_preprocess_cli.py

@@ -315,11 +315,35 @@ def test_run_from_config_uses_runtime_identity_workspace_id(monkeypatch, tmp_pat
     command.run_from_config(config)
 
     assert calls["init"]["sparse_language"] == "english"
-    assert callable(calls["init"]["candidate_evaluator"])
+    assert calls["init"]["candidate_evaluator"] is None
     assert calls["run_as_job"]["workspace_id"] == "psd-clinical"
     assert calls["run_as_job"]["input_fingerprint"] != calls["run_as_job"]["config_fingerprint"]
 
 
+@pytest.mark.parametrize("schema_version, source_type", [
+    (1, "filesystem"),
+    (1, "http"),
+    (1, "s3"),
+    (2, "http"),
+    (2, "s3"),
+])
+def test_candidate_evaluation_is_not_required_outside_v2_filesystem_corpora(
+    schema_version, source_type,
+):
+    import tht.cli.preprocess_cmd as command
+
+    cfg = SimpleNamespace(
+        evidence=SimpleNamespace(
+            schema_version=schema_version,
+            source_root=None,
+            sources=[SimpleNamespace(type=source_type)],
+        ),
+        language="it",
+    )
+
+    assert command._candidate_evaluator(cfg, vector_store=object(), embedder=object()) is None
+
+
 def test_preprocess_evidence_gc_json_is_pristine(monkeypatch, tmp_path):
     import tht.cli.preprocess_cmd as command