fix(addie): drop the business-hours gate on wg-slack-context

[bokelley]

Summary

Removes the 9–17 ET weekday window from the wg-slack-context job. The gate pattern exists for jobs that notify humans; this job's output is a PR that sits and waits for review, and the community is global — an ET window only delays the digest. The hourly tick + one-refresh-per-UTC-day dedup (from #5831) stays; the digest now lands on the first tick after 00:00 UTC, weekends included.

Test plan

• tsc --noEmit clean; precommit suite on commit (registration-only change; job logic and its 16 tests untouched).
• Prod: digest PR appears on addie/wg-slack-context within ~an hour of deploy instead of waiting for the ET window.

No changeset — Addie server work.

🤖 Generated with Claude Code

[aao-secretariat]

Drops the 9–17 ET / skip-weekends gate on wg-slack-context. Right call: the gate exists to spare humans off-hours pings, and this job's output is a PR that waits for review, not a notification — an ET window on a global community only delays the digest.

Things I checked

• No duplicate PRs. upsertFilePr targets a fixed branch and reuses the open PR (github-pr.ts:158-168), so extra ticks update the existing PR — they never open a second one. Idempotent by branch, independent of the gate.
• No runaway ticks. Interval unchanged (hourly, job-definitions.ts:395). Removing the gate raises reach from ~9/weekday to 24/day, but 23 of those short-circuit at the dedup read (wg-slack-context.ts:174-186) once the day's marker exists — two ref reads, negligible.
• Dedup is UTC, matching the new comment. today is new Date().toISOString().slice(0,10) (wg-slack-context.ts:174), and the - Generated: marker is written the same way (:142). "One refresh per UTC day, first tick after 00:00 UTC" is accurate on the wire.
• Digests aren't lost — the old boot-outside-window footgun shrinks. The comment's own concern (a 24h job that boots outside the window never running) is reduced by removing the gate, not worsened. Weekends now covered.
• No changeset is correct — server-only Addie work, no published-package or wire surface.

Follow-ups (non-blocking — file as issues)

• Twin comment left stale. wg-slack-context.ts:170-173 still reads "The scheduler's business-hours gate silently skips ticks, so this job runs on an hourly interval..." — the gate it justifies no longer exists after this PR. The PR rewrote the rationale in job-definitions.ts:386-393 but narrowed its diff to the registration file, so the runner's copy now describes a gate that's gone. Update it to mirror the new rationale.

Minor nits (non-blocking)

1. First-tick-after-00:00-UTC holds only at steady state. initialDelay: 14 minutes offsets the first post-deploy tick, so the daily-boundary claim in the rewritten comment (job-definitions.ts:391-393) is a steady-state statement. Accurate as written; optionally note the boot offset.

The unchecked prod test-plan item — digest lands within ~an hour of deploy — is the deploy-time observation, not a code path this diff can validate; non-critical, fine as a Follow-up.

LGTM. Follow-up noted.