fix: recovery sweep skips docs that already have global_events — prevents re-enqueue loop

services/scheduler/app.py

@@ -522,6 +522,9 @@ async def recover_stale_documents(pool: asyncpg.Pool, rds: aioredis.Redis) -> in
            LEFT JOIN document_company_mentions dcm ON d.id = dcm.document_id
            WHERE d.status = 'parsed'
              AND d.updated_at < NOW() - INTERVAL '1 minute' * $1
+             AND NOT EXISTS (
+                 SELECT 1 FROM global_events ge WHERE ge.source_document_id = d.id
+             )
            ORDER BY d.created_at ASC
            LIMIT 100""",
         STALE_PARSED_THRESHOLD_MINUTES,