feat(evidence): use server-side Qdrant BM25 retrieval

harness/tht/ports/vector.py

@@ -39,6 +39,8 @@ class VectorWriteRecord:
     record: VectorRecord
     embedding: list[float]
     content_hash: str
+    sparse_text: str | None = None
+    sparse_language: str | None = None
 
 
 class VectorStoreError(Exception):
@@ -74,6 +76,8 @@ class VectorStore(Protocol):
         limit: int,
         kinds: list[str] | None = None,
         metadata_filter: dict[str, object] | None = None,
+        query_text: str | None = None,
+        query_language: str | None = None,
     ) -> list[VectorHit]: ...
 
     def existing_hashes(self, collection: str, kinds: list[str]) -> dict[str, str]: ...