Fix Docker healthcheck hanging gunicorn workers

The healthcheck tried HTTPS first (curl -fsk https://...) when HTTPS
was disabled. The TLS ClientHello to a plain HTTP listener hung the
sync worker indefinitely. With 2 workers, both got stuck, blocking
all real HTTP requests.

Fix: try HTTP first, add --max-time 3 to release quickly on failure.
Compose override uses HTTP-only to match HTTPS_ENABLED=false default.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

docker/Dockerfile

@@ -65,7 +65,7 @@ WORKDIR /app/soosef
 
 EXPOSE 35811
 
-HEALTHCHECK --interval=30s --timeout=10s --start-period=15s --retries=3 \
+HEALTHCHECK --interval=30s --timeout=5s --start-period=15s --retries=3 \
-    CMD curl -fsk https://localhost:35811/ || curl -fs http://localhost:35811/ || exit 1
+    CMD curl -fs --max-time 3 http://localhost:35811/ || curl -fsk --max-time 3 https://localhost:35811/ || exit 1
 
 ENTRYPOINT ["/app/entrypoint.sh"]

docker/docker-compose.yml

@@ -15,6 +15,12 @@ services:
     volumes:
       - soosef-data:/root/.soosef
     restart: unless-stopped
+    healthcheck:
+      test: ["CMD", "curl", "-fs", "--max-time", "3", "http://localhost:35811/"]
+      interval: 30s
+      timeout: 5s
+      start_period: 15s
+      retries: 3
     deploy:
       resources:
         limits: