feat(sources): PDF-Dokumente als neuer Quellentyp pdf_document

- SOURCE_TYPE_PATTERN um pdf_document erweitert
- src/services/pdf_ingest.py: pdfplumber + Tesseract-OCR-Fallback,
  Uebersetzung nach DE+EN, ein Pool-Artikel pro PDF
- Scheduler-Job pdf_ingest laeuft im Minuten-Takt und verarbeitet
  pdf_document-Quellen mit processed_at IS NULL
- scripts/migrate_pdf_source.py: idempotente DB-Migration
  (sources.pdf_path/pdf_sha256/processed_at, articles.headline_en/content_en)
- requirements.txt: pdfplumber, pytesseract, pdf2image, Pillow

src/main.py

@@ -298,6 +298,8 @@ async def lifespan(app: FastAPI):
     orchestrator.set_ws_manager(ws_manager)
     await orchestrator.start()
 
+    from services import pdf_ingest as _pdf_ingest
+    scheduler.add_job(_pdf_ingest.run_once, "interval", minutes=1, id="pdf_ingest", max_instances=1, coalesce=True)
     scheduler.add_job(check_auto_refresh, "interval", minutes=1, id="auto_refresh")
     scheduler.add_job(cleanup_expired, "interval", hours=1, id="cleanup")
     scheduler.add_job(daily_source_health_check, "cron", hour=4, minute=0, id="source_health")