feat(worker): reclaim in-flight jobs on startup (crash-resume)

A worker crash/restart mid-pipeline left its job stuck in matching/matched/
downloading/tagging with no auto-recovery — only the manual Retry (needs_attention
only) could rescue it. Now on startup the worker resets any such orphaned job
back to 'requested' so it gets re-claimed.

Safe at startup: this fresh worker owns none of those jobs, staging is already
cleared (clear_staging_root runs first), and the pipeline redoes intake→import
idempotently with an atomic import. Terminal jobs (imported/needs_attention) are
left untouched; claimedAt/error/downloadProgress are reset. Logged on startup.

worker 218 tests / 7-skip (in-flight→requested, terminal untouched, re-claimable).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

worker/lyra_worker/claim.py

@@ -1,5 +1,28 @@
 import psycopg
 
+# States a job passes through while a worker is actively processing it. On startup this
+# fresh worker owns none of them, so any job left here was orphaned by a crash/restart.
+_IN_FLIGHT_STATES = ("matching", "matched", "downloading", "tagging")
+
+
+def reclaim_stuck_jobs(conn: psycopg.Connection) -> int:
+    """Reset jobs left mid-pipeline by a prior worker crash/restart back to 'requested' so
+    they get re-claimed. Safe to call once at startup: nothing is processing them, staging is
+    cleared beforehand, and the pipeline redoes intake→import idempotently (atomic import).
+    Returns how many were reclaimed. Terminal jobs (imported/needs_attention) are untouched."""
+    with conn.cursor() as cur:
+        cur.execute(
+            """
+            UPDATE "Job"
+            SET state = 'requested', "currentStage" = 'intake', "claimedAt" = NULL,
+                error = NULL, "downloadProgress" = 0, "updatedAt" = now()
+            WHERE state IN ('matching', 'matched', 'downloading', 'tagging')
+            """
+        )
+        n = cur.rowcount
+    conn.commit()
+    return n
+
 
 def claim_next(conn: psycopg.Connection) -> str | None:
     """Atomically claim the oldest queued job. Returns its id or None."""

worker/lyra_worker/main.py

@@ -5,7 +5,7 @@ import uuid
 
 import psycopg
 
-from lyra_worker.claim import claim_next
+from lyra_worker.claim import claim_next, reclaim_stuck_jobs
 from lyra_worker.config import get_config
 from lyra_worker.crypto import secret_key_problem
 from lyra_worker.db import wait_for_db
@@ -207,6 +207,9 @@ def run_forever() -> None:
     _require_secret_key()
     conn = wait_for_db()
     clear_staging_root(STAGING_ROOT)  # sweep any staging dirs orphaned by a prior crash
+    reclaimed = reclaim_stuck_jobs(conn)  # requeue jobs a prior crash left mid-pipeline
+    if reclaimed:
+        print(f"worker: reclaimed {reclaimed} in-flight job(s) from a prior crash", flush=True)
     adapters = build_adapters(get_config(conn))
     resolver = build_resolver()
     tagger = build_tagger()