[codex] Add Binance dispatch retry guard

[Pigbibi]

Summary

• add bounded retry/backoff to the VPS Binance runtime dispatch guard
• retry only transient curl failures and GitHub 000/500/502/503/504 dispatch responses
• document the retry behavior and add a static workflow regression test

Root Cause

The 2026-06-02 16:00 UTC alert happened before a Runtime Actions run was created: GitHub returned HTTP 500 from the workflow dispatch API. The existing guard treated every non-204 response as final, so one transient API error could skip that hour.

Validation

• git diff --check
• python -m unittest tests.test_dispatch_guard_workflow -v
• extracted generated dispatch-runtime.sh from the workflow and ran bash -n

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: d972dbc8e8

ℹ️ About Codex in GitHub

Your team has set up Codex to 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 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Avoid retrying ambiguous dispatch successes

When GitHub creates the workflow_dispatch run but the client receives a transient 500/502/503/504, retrying this POST sends another dispatch because the request has no idempotency key; I checked .github/workflows/main.yml and it does not define a concurrency guard, so this can start overlapping live Runtime jobs after a single ambiguous server-side success, contrary to the runbook's scheduler-overlap warning. Consider verifying whether a run was created before retrying, or adding workflow-level concurrency before treating these statuses as retryable.

Useful? React with 👍 / 👎.