Offload project scaffolding writes from the event loop

[groupthinking]

Move all ProjectCodeGenerator scaffolding filesystem calls off the event loop into a single batched worker hop per generator and keep cleanup atomic under cancellation.

[linear-code]

GRV-255

[github-actions]

👋 Thanks for opening this issue! It has been automatically labeled for triage.

[coderabbitai]

⚠️ Possible Duplicate Issue(s)

• perf: project scaffolding disk writes block the event loop in ProjectCodeGenerator #1250

---

📝 Issue Planner

_{Check the box below or use the @coderabbitai plan command to generate an implementation plan and prompts that you can use with your favorite coding assistant.}

• Create Plan

---

🧪 Issue enrichment is currently in open beta.

You can configure auto-planning by selecting labels in the issue_enrichment configuration.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord!

[groupthinking]

Resolved on `main` by #1251 (merged) — closing as triage hygiene.

Verified in current `origin/main`, `src/youtube_extension/backend/code_generator.py`:

• L375 — `await asyncio.to_thread(_apply_write_plan, plan)`
• L429 — `await asyncio.to_thread(_apply_write_plan, plan)`
• L479 — `await asyncio.to_thread(_apply_write_plan, plan)`

All three scaffolding generators build an in-memory write plan and flush it in a single worker-thread hop, so `mkdir`/`write_text` no longer block the loop per file.

Note: the placeholder-endpoint concern that #1251 deliberately scoped out (it needed byte-identical generated output to prove the refactor was behaviour-preserving) is tracked separately in #1257 and addressed in #1289.