Fix: Remove VLM pre-cache to prevent build timeout

Dockerfile

@@ -40,10 +40,8 @@ RUN mkdir -p /app/models \
 COPY requirements.txt .
 RUN pip install --no-cache-dir -r requirements.txt
 
-# Pre-download the VLM model (so it's cached in the image)
-RUN python -c "from transformers import AutoProcessor, AutoModelForVision2Seq; \
-    AutoProcessor.from_pretrained('HuggingFaceTB/SmolVLM2-256M-Video-Instruct'); \
-    AutoModelForVision2Seq.from_pretrained('HuggingFaceTB/SmolVLM2-256M-Video-Instruct')"
+# Note: SmolVLM model downloads on first use (~500MB)
+# This avoids build timeouts on free tier
 
 # Copy Backend Code
 COPY backend ./backend