Seed leases alternate between priorities; article relay can be off

Half the live leases go to higher-priority ranges, the rest to the others,
so a slow recent gap runs beside older history. With no article
connections the article relay answers 503 instead of waiting forever.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

app/services/SeedOrchestrator.ts

@@ -73,12 +73,21 @@ class SeedOrchestrator {
        WHERE status = 'leased' AND heartbeat_at < now() - make_interval(secs => ?) AND attempts >= ?`,
       [LEASE_TTL_SECONDS, MAX_ATTEMPTS]
     )
+    // Priorities take turns: while ranges of a higher priority are leased for less than half of
+    // the work in progress they go first, otherwise the rest does (a 2026 gap fetches ~5x slower
+    // than older history; LOG.md 2026-10-06 (10)).
+    const { rows: mix } = await db.rawQuery(
+      `SELECT count(*) FILTER (WHERE priority > 0) AS high, count(*) AS all FROM seed_ranges
+       WHERE status = 'leased' AND heartbeat_at >= now() - make_interval(secs => ?)`,
+      [LEASE_TTL_SECONDS]
+    )
+    const order = Number(mix[0].high) * 2 < Number(mix[0].all) + 1 ? 'priority DESC, id' : 'priority ASC, id'
     const { rows } = await db.rawQuery(
       `UPDATE seed_ranges SET status = 'leased', node_id = ?, leased_at = now(), heartbeat_at = now(), attempts = attempts + 1
        WHERE id = (
          SELECT id FROM seed_ranges
          WHERE status = 'queued' OR (status = 'leased' AND heartbeat_at < now() - make_interval(secs => ?))
-         ORDER BY priority DESC, id LIMIT 1 FOR UPDATE SKIP LOCKED)
+         ORDER BY ${order} LIMIT 1 FOR UPDATE SKIP LOCKED)
        RETURNING id, group_name, next_id, end_id`,
       [nodeId, LEASE_TTL_SECONDS]
     )