Summary

• AppInsights flush() now runs in a daemon=True thread with a 5s timeout
• If the Azure endpoint is unreachable (e.g. Automation Suite / CI with restricted egress), the flush is abandoned after 5s instead of blocking indefinitely
• Bumps uipath version to 2.10.40

Problem

In Automation Suite environments, coded agent eval runs were getting stuck in Running state forever. The root cause:

1. The Python process completes the eval (PUT to update eval run succeeds)
2. AppInsightsEventClient.flush() is called (via finally block or atexit handler)
3. AS/CI environments block outbound access to dc.applicationinsights.azure.com
4. urlopen() times out → _DiagnosticSender re-queues the items → SynchronousQueue.flush() retries in a loop
5. Process never exits → Orchestrator never receives the exit signal → job stays "Running"

Alpha works fine because it has outbound internet access and the flush completes in <1s.

Fix

thread = threading.Thread(target=_do_flush, daemon=True)
thread.start()
thread.join(timeout=timeout)  # default 5s
if thread.is_alive():
    _logger.warning("AppInsights flush: timed out after %.1fs, abandoning send", timeout)

Using daemon=True ensures the thread is abandoned on process exit even if join() somehow doesn't return in time.

Test plan

• Existing flush tests pass (8 passed)
• Verify eval runs complete in AS/CI environments (no longer stuck in Running)
• Verify alpha behaviour unchanged (flush completes within 5s, no timeout warning)