fix: re-enable intermediate flush for all attempts, add rollback note

The is_final_attempt guard disabled intermediate flush for the common
case (first attempt succeeds, which is 99%+ of turns). Retries only
fire on context-too-long errors with events_yielded==0, meaning the
stream barely started and flush threshold was almost certainly not
reached. Keep flush always enabled and document the theoretical edge
case.

autogpt_platform/backend/backend/copilot/sdk/service.py

@@ -266,7 +266,6 @@ class _RetryState:
     adapter: SDKResponseAdapter
     transcript_builder: TranscriptBuilder
     usage: _TokenUsage
-    is_final_attempt: bool = False
 
 
 @dataclass
@@ -1548,11 +1547,9 @@ async def _run_stream_attempt(
             # --- Intermediate persistence ---
             # Flush session messages to DB periodically so page reloads
             # show progress during long-running turns.
-            # Only flush on the final attempt to avoid persisting messages
-            # that may be rolled back if a transient error triggers a retry.
             _msgs_since_flush += 1
             now = time.monotonic()
-            if state.is_final_attempt and (
+            if (
                 _msgs_since_flush >= _FLUSH_MESSAGE_THRESHOLD
                 or (now - _last_flush_time) >= _FLUSH_INTERVAL_SECONDS
             ):
@@ -2043,10 +2040,6 @@ async def stream_chat_completion_sdk(
         )
 
         for attempt in range(_MAX_STREAM_ATTEMPTS):
-            # Enable intermediate DB flushing only on the final attempt so
-            # earlier attempts (which may be rolled back) don't persist
-            # messages that are later removed from session.messages.
-            state.is_final_attempt = attempt == _MAX_STREAM_ATTEMPTS - 1
             # Clear any stale stash signal from the previous attempt so
             # wait_for_stash() doesn't fire prematurely on a leftover event.
             reset_stash_event()
@@ -2172,6 +2165,11 @@ async def stream_chat_completion_sdk(
                     exc_info=True,
                 )
                 session.messages = session.messages[:pre_attempt_msg_count]
+                # Note: if intermediate flushes persisted messages from this
+                # attempt, those orphans remain in DB.  In practice this is
+                # unlikely because retries only fire on context-too-long errors
+                # with events_yielded==0 (meaning the stream barely started and
+                # the flush threshold was almost certainly not reached).
                 # Cancel any pre-launched tasks from the failed attempt so they
                 # don't continue executing against the rolled-back session.
                 await cancel_pending_tool_tasks()