test: cover context-engine-active recall bypass

[haosenwang1018]

Summary

• add a real-server plugin test for the OpenClaw compatibility fix path
• cover the case where the OpenViking context-engine is instantiated before before_prompt_build
• assert that hook recall is skipped in that case while assemble() still returns OpenViking-backed context

Testing

• git diff --check

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[github-actions]

PR Reviewer Guide 🔍

Here are some key observations to aid the review process:

⏱️ Estimated effort to review: 2 🔵🔵⚪⚪⚪

🏅 Score: 95

🧪 PR contains tests

🔒 No security concerns identified

✅ No TODO sections

🔀 No multiple PR themes

⚡ No major issues detected

[github-actions]

PR Code Suggestions ✨

No code suggestions found for the PR.

[qin-ctx]

I think we should distinguish recall and assemble here, since they do not overlap in responsibility:

• before_prompt_build recall is for long-term memory retrieval. It queries OpenViking memory/search with the current query and prepends matched user/agent memories to the prompt.
• context-engine assemble() is for current-session short-term context reconstruction. It reads OpenViking session context and rebuilds archive summaries / active messages into OpenClaw-consumable messages.

So assemble() is not a replacement for recall, and this PR should not describe them as the same capability.
Thanks for your contribution !