Load RAG index eagerly at container startup instead of on first request

modules/rag.py

@@ -1,6 +1,7 @@
 """
-Loads the prebuilt local RAG index (rag_index/, built by scripts/build_index.py)
-and retrieves anonymized context chunks. Never crawls or chunks anything itself.
+Loads the prebuilt local RAG index (rag_index/, built by
+scripts/build_index_from_crawler.py) and retrieves anonymized context chunks.
+Never crawls or chunks anything itself.
 
 Retrieved chunks carry no title/author/URL/domain — the index physically does not
 store that metadata — so callers can only ever label them "Source 1", "Source 2", ...
@@ -71,6 +72,11 @@ def is_available() -> bool:
     return _collection is not None
 
 
+def chunk_count() -> int:
+    _load()
+    return _collection.count() if _collection is not None else 0
+
+
 def retrieve(query: str, top_k: int = 8, category: str | list[str] | None = None) -> list[str]:
     """Returns ["Source 1: <text>", "Source 2: <text>", ...], or [] if unavailable.