feat: split backend into lib/ modules + add squid.wtf (Qobuz) download backend

- backend/lib/config.js  — CONFIG from env (adds DOWNLOAD_BACKEND, SQUID_QUALITY)
- backend/lib/jobs.js    — jobs Map, updateJob(), log()
- backend/lib/metadata.js — iTunes, artwork, lyrics, sanitize helpers
- backend/lib/ytdlp.js   — yt-dlp download cascade + shared embedMetadata()
- backend/lib/squid.js   — ALTCHA PoW solver, Qobuz search/download via squid.wtf
- backend/lib/downloader.js — backend selector: ytdlp | squid | auto (squid-first + fallback)
- novasonic-server.js thinned to Express routes only, imports from lib/
- docker-compose: mount full backend/ dir, install deps outside read-only volume
- .env.example: document DOWNLOAD_BACKEND and SQUID_QUALITY

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

.env.example

@@ -5,9 +5,20 @@ NAVIDROME_URL=https://music.yourdomain.com
 BACKEND_URL=https://novasonic-api.yourdomain.com
 
 # Absolute path to your Navidrome music library on the host
-# This exact path is used inside the container — no subdirectory is added
 MUSIC_PATH=/srv/navidrome/music
 
-AUDIO_FMT=mp3          # mp3 | flac | opus | best
-AUDIO_QUALITY=0        # 0 = best quality
+AUDIO_FMT=mp3          # mp3 | flac | opus | best  (used by yt-dlp)
+AUDIO_QUALITY=0        # 0 = best quality           (used by yt-dlp)
 MAX_PARALLEL=2
+
+# Download backend: ytdlp | squid | auto
+#   ytdlp  - always use yt-dlp
+#   squid  - always use squid.wtf (Qobuz)
+#   auto   - try squid.wtf first, fall back to yt-dlp if not found (default)
+DOWNLOAD_BACKEND=auto
+
+# Quality for squid.wtf / Qobuz downloads
+#   mp3    - MP3 320 kbps
+#   flac   - FLAC 16-bit / 44.1 kHz
+#   hi-res - FLAC 24-bit / 96 kHz
+SQUID_QUALITY=mp3