fix: Dockerfile (resolve RUNTIME_ERROR / gradio launch)

Dockerfile

@@ -6,6 +6,10 @@ ENV DEBIAN_FRONTEND=noninteractive \
     JAVA_HOME=/usr/lib/jvm/java-21-openjdk-amd64 \
     GRADLE_USER_HOME=/tmp/.gradle \
     HF_HOME=/tmp/hf_cache \
+    HF_HUB_CACHE=/tmp/hf_cache \
+    PORT=7860 \
+    GRADIO_SERVER_NAME=0.0.0.0 \
+    GRADIO_SERVER_PORT=7860 \
     PATH=/opt/android-sdk/cmdline-tools/latest/bin:/opt/android-sdk/platform-tools:/usr/lib/jvm/java-21-openjdk-amd64/bin:/opt/gradle-8.7/bin:$PATH
 
 # Base tooling
@@ -43,7 +47,17 @@ COPY --chown=user:user requirements.txt .
 RUN pip install --no-cache-dir --upgrade pip && \
     pip install --no-cache-dir -r requirements.txt
 
+# Pre-download the GGUF model at build time so the first user request does not
+# have to wait for (and risk timing out on) a ~1GB download at runtime. This is
+# the main reason the Space booted into RUNTIME_ERROR before the UI loaded.
+RUN python -c "from huggingface_hub import hf_hub_download; \
+hf_hub_download(repo_id='Qwen/Qwen2.5-Coder-1.5B-Instruct-GGUF', \
+filename='qwen2.5-coder-1.5b-instruct-q4_k_m.gguf')" || \
+echo 'model pre-download skipped (will download at runtime)'
+
 COPY --chown=user:user . .
 
 EXPOSE 7860
-CMD ["python", "app.py"]
+
+# Use exec form so signals propagate; app.py reads $PORT.
+CMD ["python", "app.py"]