Fix research pipeline budget tracking

src/nsct/crawler/manager.py

@@ -167,7 +167,10 @@ class CrawlerManager:
     @staticmethod
     def _error_doc(url: str, error: str) -> NormalizedDocument:
         """Create a NormalizedDocument representing an error."""
-        doc = NormalizedDocument(
+        # Keep the error result structurally identical to a successfully
+        # normalized document.  ``from_text`` hashes the (empty) content, so
+        # downstream consumers can safely rely on content_hash being present.
+        doc = NormalizedDocument.from_text(
             url=url,
             text="",
             title="",
@@ -176,7 +179,6 @@ class CrawlerManager:
                 "error": error,
                 "content_type": "error",
             },
-            content_hash="",
             extraction_tool="",
         )
         return doc