core: make idle resume decisions atomic

[winston-openai]

Summary

• make idle resume config comparison and shutdown one atomic core decision, so a completed settings update cannot trigger a stale cold replacement
• serialize runtime user-config refreshes with the same submission/activity lifecycle and expose stable in-memory agent subtree IDs
• cover the public queued trigger-turn transition requested in the lower-stack review

Validation

• focused core tests for atomic reuse/shutdown outcomes and runtime-config refresh serialization
• focused core tests for unloaded intermediate agent ancestry, failed idle delivery, and cancellation cleanup
• public test_codex coverage for trigger-turn mail arriving at turn completion

Stack

• based on app-server: own thread listener teardown #31360

[winston-openai]

@codex review

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. Chef's kiss.

Reviewed commit: ee25fa1487

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[winston-openai]

@codex review

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. What shall we delve into next?

Reviewed commit: 69309d111e

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".