feat: ElevenLabs-Voice-Liste + Fehlerdetails in Audio-Batch-Ergebnissen

- GET /api/tts-settings/voices/available listet die Account-Stimmen
  (Grundlage für Voice-Auswahl im CMT statt Freitext-IDs)
- Audio-Batch/-Fill-Fehler enthalten jetzt das ElevenLabs-Detail
  (z.B. voice_not_found) statt nur 'ElevenLabs error'

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

src/routes/pipeline.js

@@ -4,6 +4,7 @@ const { query } = require('../db');
 const { LANGS } = require('../lib/translate');
 const { loadPairContext, computeReadiness, loadPairContent } = require('../lib/pairContent');
 const { enqueue, loadPairs, collectAudioUnits, generateWithBackoff, translatePair } = require('../lib/pipeline');
+const { describeError } = require('./audios');
 const { PLACEHOLDER_RE } = require('../lib/placeholders');
 
 // ── Objekt-Wort-Erkennung in Sätzen (für die manuelle Zuweisung beim Review) ──
@@ -233,7 +234,7 @@ router.post('/picture/:id/audio-fill', async (req, res, next) => {
       try { await generateWithBackoff(u); result.generated++; }
       catch (err) {
         result.failed++;
-        result.errors.push({ source: `${u.source_table}/${u.source_field}/${u.language}`, error: err.message });
+        result.errors.push({ source: `${u.source_table}/${u.source_field}/${u.language}`, error: describeError(err) });
       }
     }
     res.json(result);