Station signup: jobs/station-signup-attempt-prune for 30-day attempt-log retention

[jakebromberg]

Problem

station_signup_attempt grows without bound. Rows are retained 30 days for audit (they answer "who revealed the code?" and "what did the attack look like?"), then need deleting.

Desired end state

A jobs/station-signup-attempt-prune workspace deleting attempt rows older than 30 days, consuming pruneSignupAttempts from the lifecycle module.

This is not one file. jobs/* is an npm workspace (package.json:67-71). The nearest shape, jobs/flowsheet-ghost-row-sweep/, carries job.ts, package.json, tsconfig.json, tsup.config.ts, and a README, plus a root Dockerfile.flowsheet-ghost-row-sweep and deploy/cron wiring. Follow it.

Cron registration

Register in docs/ops-cron-scheduling.md under "Excluded / DB-only (verified — don't try to 'space' these)" (line 46), beside metadata-no-match-digest — not in the slot table. That table exists to space LML-hitting crons, and a pure-Postgres prune is not a scheduling constraint.

Acceptance criteria

• Workspace with the full jobs/* footprint
• Root Dockerfile + deploy wiring
• Registered in the DB-only section of docs/ops-cron-scheduling.md
• Deletes only rows older than 30 days; a test pins the boundary

Context

Plan: ~/Downloads/wxyc-station-passcode-signup-plan.md.

Related

Blocked by the lifecycle module. Split from the review job because the failure consequences differ: a failed prune means a larger table, a failed review means a silent queue and a security gap, and those want separate alerting.

Blocked by

• Station signup: station-passcode.ts lifecycle module (encrypt, verify, rotate, cooldown) #2359