[codex] Add Python SDK goal turns

[aibrahim-oai]

Summary

• expose goal=True on synchronous and asynchronous Python run and turn
• start persisted goals atomically through turn/start while preserving input and turn options
• present goal continuations as one logical SDK turn with stable IDs, aggregated results, and rollover-aware controls
• regenerate app-server schemas and Python models, with API docs, examples, and focused coverage

Validation

• just fix -p codex-core -p codex-app-server -p codex-extension-api -p codex-goal-extension
• just write-app-server-schema
• uv run --frozen --no-sync python scripts/update_sdk_artifacts.py generate-types
• Python Ruff formatting and checks
• just fmt

Tests were not run locally; CI will run the added Rust and Python coverage.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 851a85b369

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]

Split goal-turn work into reviewable stages

The AGENTS.md “Change size guidance (800 lines)” rule asks non-mechanical changes to stay under 800 lines, or under 500 lines for complex logic. This diff is about 2.9k lines across app-server/core/extensions and the Python SDK, so review is too broad. Please split the core/app-server goal-turn plumbing from the Python SDK/docs/tests stage so each part can be reviewed independently.

Useful? React with 👍 / 👎.

[aibrahim-oai]

Superseded by #26953, whose head branch is hosted directly in openai/codex.