ci(release): check for a nightly every three hours

[rynfar]

What Changed

The nightly schedule in release.yml goes from 0 9 * * * (daily) back to 0 */3 * * * (every three hours), with the comment rewritten to say why. One line of YAML plus its comment.

Why

61ba5899e throttled the inherited three-hour cadence to once a day because "each rebuild is billable" on Blacksmith. After #99 the repository is public and every job runs on GitHub-hosted standard runners, where minutes are free, so that reason is gone.

What keeps this from costing more:

• check_changes still runs first and skips the whole release when pylon has not moved since the last nightly tag, so an idle three-hour slot is one ~20 s Linux job.
• The nightly tag already embeds the run number (vX.Y.Z-nightly.YYYYMMDD.<run>), so several builds on one day cannot collide.
• The release-nightly concurrency group with cancel-in-progress means a new scheduled run supersedes a stale in-flight one rather than stacking.
• The only per-build services with quotas are Azure Trusted Signing (Windows) and the pylon-releases asset feed; at a maximum of eight builds on an active day both stay far inside their limits.

docs/operations/release.md already documents "scheduled check every three hours" in two places — it was never updated for the daily throttle — so this change brings the workflow back in line with the runbook rather than requiring a doc edit.

Checklist

• This PR is small and focused
• I explained what changed and why
• I included before/after screenshots for any UI changes (no UI change)
• I included a video for animation/interaction changes (no motion change)

Model: Claude Fable 5 · Harness: Claude Code

---

^{Need help on this PR? Tag @codesmith-bot with what you need. Autofix is disabled.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.3 KiB	13.3 KiB	+9 B (+0.1%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+6 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+3 B (+0.0%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	55.6 KiB	55.6 KiB	0 B (0.0%)	66.4 KiB	✅
Codex	Live turn messages	10	10	0 (0.0%)	21	✅
Claude	Total thread wire	13.3 KiB	13.3 KiB	+15 B (+0.1%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	0 B (0.0%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+15 B (+0.2%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	56.4 KiB	56.4 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: 4880682 · PR result: 99cd4c8 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 109.4 KiB
• Claude decoded thread snapshot: 110.0 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.