Resume an interrupted bulk submission

ws.submit() journals every outcome to .battinfo/submit-journal.jsonl as it happens. If a 10k-record submission dies at record 7,000, just run it again:

ws.submit()               # re-run after the interruption
# already-submitted records print:
#   <title>  [already submitted — skipped; pass resume=False to re-send]

• Succeeded records are skipped locally (status: skipped_journal) — no re-POST round-trips.

• Failed records retry automatically on the next run.

• Changed content re-submits — the journal keys on a content hash, so an edited record never hides behind an old success.

• ws.submit(resume=False) bypasses the journal entirely; the registry dedups identical re-submissions server-side anyway.

Transient registry failures (429/5xx, timeouts) are retried with exponential backoff within each submission — resuming is for interruptions on your side.