Add live stream-json idle events and contract docs

[konard]

Summary

• Emit a public session_idle event when a session reaches a turn boundary and map it to Claude stream-json as idle.
• Use the shared bus event output path from continuous stdin mode so live drivers get the same session, permission, error, and idle events as single-message JSON mode.
• Document the live stream-json contract for hive-mind style drivers: stdin frames, resume/no-fork behavior, user-frame replay acknowledgements, interrupts, queueing, and idle boundaries.
• Add a changeset for the patch release.

Reproduction

Before this change, session.idle only resolved the internal wait promise. A driver using --input-format stream-json --output-format stream-json could see user-frame replay and result events, but had no explicit public turn-boundary event to know when queued input could safely be flushed.

The new js/tests/event-handler.js reproduces that gap by publishing SessionStatus.Event.Idle and asserting a public session_idle output is produced.

Tests

• cd js && bun test ./tests/event-handler.js ./tests/json-standard-unit.js ./tests/input-queue.js
• cd js && npm run check
• cd js && npm test (600 pass, 4 todo)
• node scripts/validate-changeset.mjs

Fixes #273

[konard]

Working session summary

Implemented and pushed the fix to issue-273-d5676a1f4939.

PR 274 is ready for review: #274

What changed:

• Added public session_idle output and Claude stream-json idle mapping.
• Shared the bus event output path with continuous stdin mode.
• Documented live stream-json resume, replay, interrupt, queueing, and idle semantics.
• Added regression coverage and a patch changeset.
• Removed the generated .gitkeep placeholder.

Verification:

• cd js && bun test ./tests/event-handler.js ./tests/json-standard-unit.js ./tests/input-queue.js
• cd js && npm run check
• cd js && npm test 600 pass, 4 todo
• node scripts/validate-changeset.mjs
• GitHub CI run 28671028299 passed on commit f5c3a3c.

---

This summary was automatically extracted from the AI working session output.

[konard]

🤖 Solution Draft Log

This log file contains the complete execution trace of the AI solution draft process.

💰 Cost estimation:

• Model: GPT-5.5
• Provider: OpenAI
• Public pricing estimate: $13.333963

📊 Context and tokens usage:

• 426.8K / 200K (213%) input tokens, 31.9K / 128K (25%) output tokens

Total: (426.8K + 7.6M cached) input tokens, 31.9K output tokens, $13.333963 cost

🤖 Models used:

• Tool: OpenAI Codex
• Requested: gpt-5.5
• Thinking level: max (~31999 tokens)
• Model: GPT-5.5 (gpt-5.5)

📎 Log file uploaded as Gist (5263KB)

• View complete solution draft log

---

Now working session is ended, feel free to review and add any feedback on the solution draft.

[konard]

🎉 Auto-merged

This pull request has been automatically merged by hive-mind.

• All CI checks have passed

---

Auto-merged by hive-mind with --auto-merge flag