Job priority queue: high-priority email tasks shouldn't wait behind background autoresearch

[FrederikHandberg]

Problem

An email from Frederik competes equally with a background autoresearch loop that's been running for 45+ minutes. Both slots can be filled with long low-priority work, and critical tasks wait in retry queue with no ETA.

No priority field exists in JobRequest. No preemption mechanism.

Fix

1. Add priority: high | normal | low field to JobRequest schema in listen server
2. Email agent sets priority: high for emails from frederik@tutoria.ac
3. Scheduled background tasks (autoresearch, morning report) set priority: low
4. Worker processes high-priority jobs first when deciding which queued job to start next
5. Stretch: if high-priority job arrives and both slots are low/normal, pause the lowest-priority job at next safe checkpoint (after YAML write)

Acceptance Criteria

• priority field in JobRequest schema
• Email agent sets correct priority based on sender
• Worker respects priority order when picking next job from queue
• Dashboard shows priority in job list
• High-priority job starts within 60 seconds even when queue has lower-priority jobs

Context

From Jensen limitation audit (2026-03-22). Workstream: WS-001.