Add presenter view, stats, lyrics worker, watch party with voice chat, timestamp sharing, soundbites, notes, transcript search, PiP, EQ, sleep fade, gestures and external players; fix the crossfade end-of-song race

scripts/lyrics/Dockerfile

@@ -0,0 +1,15 @@
+# Lyrics worker — transcribes saved songs that have no lyrics yet, one at a
+# time, with faster-whisper on CPU (no API keys). Talks to the ytplayer
+# service over the compose network only; see scripts/lyrics/auto_lyrics.py.
+FROM python:3.12-slim
+ENV PYTHONUNBUFFERED=1 \
+    HF_HOME=/models \
+    PIP_NO_CACHE_DIR=1
+WORKDIR /app
+COPY requirements.txt .
+RUN pip install -r requirements.txt
+COPY auto_lyrics.py .
+# Model weights (downloaded on first use) and the worker's memory of
+# instrumentals/failures live on volumes, so rebuilds don't re-download.
+VOLUME ["/models", "/data"]
+CMD ["sh", "-c", "exec python auto_lyrics.py --missing --watch ${WATCH_SECONDS:-300} --state /data/state.json --model ${WHISPER_MODEL:-large-v3-turbo} --threads ${WHISPER_THREADS:-2}"]