test(usage): close round-5 nits (empty doc_types, consistency tidy-ups)

Round-5 claude-review (merge-ready; all nits):

- 🟡 Added test_empty_doc_types_normalizes_to_null pinning doc_types=[] → None
  in record_search_usage metadata (matches the None case).
- 🟡 record_search_usage docstring now notes nc_semantic_search_answer always
  meters with doc_types=None (it exposes no doc_types parameter).
- 🟢 BM25HybridSearchAlgorithm.__init__ now sets query_embedding /
  query_token_count alongside _embedded_query, so all three cache fields are
  instance attributes from construction (was relying on the class-level
  SearchAlgorithm defaults).
- 🟢 Ollama embed_batch_with_usage caches _dimension inline (mirrors
  OpenAI/Mistral), so the dimension is set via any embed path.
- 🟢 record_indexing_usage documents the independent-record / partial-failure
  semantics under SUM aggregation.

Deck #67.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

nextcloud_mcp_server/server/semantic.py

@@ -70,7 +70,9 @@ async def record_search_usage(
     estimated) — the unit upstream providers bill on, and the same metric the
     indexing path records for chunk embeddings (Deck #67). ``nc_semantic_search``
     and ``nc_semantic_search_answer`` (which reuses it) both flow through here —
-    do not add a second hook.
+    do not add a second hook. ``nc_semantic_search_answer`` exposes no
+    ``doc_types`` parameter, so its searches always meter with
+    ``doc_types=None``.
 
     Best-effort and flag-gated: a metering failure is logged and never breaks
     the search. Unlike the indexing path's chunk-count guard, a 0-token query is