fix: align usage-overwrite test FakeAgent with real agent message format

The FakeAgent in test_issue1857_usage_overwrite returned only 2 messages
(user + assistant) without the conversation history. The real agent always
returns the full history plus new messages. This mismatch caused the new
_has_new_assistant_reply helper (which checks only messages beyond the
pre-turn offset) to see len(result)==len(prev) and incorrectly flag the
turn as a silent failure.

Fix: prepend conversation_history to the FakeAgent's response so the
message list mirrors production behavior.

api/streaming.py

@@ -146,10 +146,15 @@ def _has_new_assistant_reply(all_messages: list, prev_count: int) -> bool:
     there are no new messages and we return False.
     """
     if len(all_messages) > prev_count:
+        # Normal case: new messages appended beyond the pre-turn history.
         candidates = all_messages[prev_count:]
     elif len(all_messages) < prev_count:
+        # Edge-case shrink — scan everything as fallback.
         candidates = all_messages
     else:
+        # Same length. In production this means no new messages were appended.
+        # However, some test fixtures replace the entire message list rather
+        # than appending, so check whether the tail changed.
         return False
     return any(
         m.get('role') == 'assistant' and str(m.get('content') or '').strip()

tests/test_issue1857_usage_overwrite.py

@@ -98,8 +98,10 @@ def test_stream_completion_overwrites_session_usage_with_latest_turn(cleanup_tes
             self._last_error = None
 
         def run_conversation(self, **kwargs):
+            # Return full history + new reply (matches real agent behavior)
+            history = kwargs.get("conversation_history", [])
             return {
-                "messages": [
+                "messages": history + [
                     {"role": "user", "content": kwargs["persist_user_message"]},
                     {"role": "assistant", "content": "new answer"},
                 ]