fix: resolve flaky test_nested_subagents_create_recursive_toolparts

[Million-mo]

Problem

PR #115 的 Unit tests CI 挂了——test_nested_subagents_create_recursive_toolparts 在 CI 慢速环境下死锁，触发 --timeout=60 后 rerun 也卡住，最终 10 分钟 job 超时被 cancel。

Root Cause

Depth=2 spawn event 在 parent consumer 还没完全启动时就发布了，事件丢失导致 assertion 永远等不到，测试挂起。

Fix

在发布 depth=2 spawn event 之前加 await asyncio.sleep(0.2)，确保 parent consumer 完全启动。

本地验证 3 次全部 < 1s 通过，无 rerun。

Note: 这是已有的 flaky test，与 OpenAICompatibleModel 无关。fix 在 #116 squash merge 时丢失了，这里单独补上。

[gemini-code-assist]

Code Review

This pull request adds a hardcoded sleep to prevent a race condition in a subagent events test. The reviewer correctly points out that using a hardcoded sleep is an anti-pattern and suggests replacing it with a polling loop to wait for the parent consumer to initialize, ensuring both speed and robustness.

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.

[gemini-code-assist]

Using a hardcoded asyncio.sleep(0.2) to resolve race conditions in tests is an anti-pattern. It can still be flaky on extremely slow CI runners, while unnecessarily slowing down the test suite on fast local machines.

Instead, we can use a polling loop with a timeout to wait for the parent consumer to be registered in integration._contexts, followed by a very brief sleep to allow the subscription to complete. This is both faster and more robust.

Suggested change

	# Extra sleep to ensure parent consumer is fully started before publishing
	# depth=2 event — without this, the event can be missed on slow CI runners.
	await asyncio.sleep(0.2)
	# Wait for parent consumer to initialize
	for _ in range(50):
	if parent_id in integration._contexts:
	break
	await asyncio.sleep(0.01)
	await asyncio.sleep(0.05)