fix: six P2 correctness/robustness cleanups from the SDK review

- legacy call_llm: open AsyncOpenAI via `async with` so the client (and
  its HTTP connection pool) is closed instead of leaked.
- LocalBackend.add_document: fail fast with CollectionNotFoundError when
  the collection doesn't exist, before the expensive parse + LLM index
  (previously the missing FK only tripped at save time, after paying for
  the LLM work). Also raise builtin FileNotFoundError for a missing path
  instead of FileTypeError (which now means only "unsupported extension").
- Collection.query(doc_ids=None): the empty-collection guard now always
  runs — previously it was skipped once PAGEINDEX_EXPERIMENTAL_MULTIDOC
  was set. A single list_documents call serves both the guard and the
  multi-doc warning (no separate call just to decide whether to warn).
- CloudBackend.query_stream: on early consumer break / GeneratorExit,
  signal the background SSE thread to stop and force-close the response,
  so it no longer drains the whole stream in the background.

Adds regression tests for each (client closed, fail-fast on unknown
collection, FileNotFoundError, empty-check under the multidoc env flag,
single list call, early-break thread stop).

Claude-Session: https://claude.ai/code/session_01Kx5DgKbhK1N8autqXH8SmS

tests/test_collection.py

@@ -94,3 +94,26 @@ def test_query_accepts_str_doc_id(col):
 def test_query_rejects_empty_list(col):
     with pytest.raises(ValueError, match="cannot be empty"):
         col.query("what?", doc_ids=[])
+
+
+def test_empty_collection_check_runs_even_when_multidoc_acked(monkeypatch):
+    from unittest.mock import MagicMock
+    from pageindex.collection import Collection
+    monkeypatch.setenv("PAGEINDEX_EXPERIMENTAL_MULTIDOC", "1")
+    backend = MagicMock()
+    backend.list_documents.return_value = []
+    col = Collection(name="papers", backend=backend)
+    with pytest.raises(ValueError, match="empty"):
+        col.query("q")  # doc_ids=None, collection empty -> must still raise
+
+
+def test_whole_collection_query_lists_documents_once(monkeypatch):
+    from unittest.mock import MagicMock
+    from pageindex.collection import Collection
+    monkeypatch.setenv("PAGEINDEX_EXPERIMENTAL_MULTIDOC", "1")  # silence warning path
+    backend = MagicMock()
+    backend.list_documents.return_value = [{"doc_id": "d1"}, {"doc_id": "d2"}]
+    backend.query.return_value = "ans"
+    col = Collection(name="papers", backend=backend)
+    col.query("q")
+    assert backend.list_documents.call_count == 1  # single call at the collection layer