fix: rename ollama→ollama-cpu, fix llama-cpp-hermes YAML indentation

- Rename ollama service to ollama-cpu (CPU-only for bge-m3 embeddings)
- Fix llama-cpp-hermes indented under networks instead of as top-level service
- Update hermes OLLAMA_HOST to point to ollama-cpu

Part of PR #54 GPU/ROCm refactor

ai/compose.yml

@@ -38,7 +38,7 @@ services:
     # Gateway run enables the internal API server on port 8642
     command: gateway run
     environment:
-      - OLLAMA_HOST=http://ollama:11434
+      - OLLAMA_HOST=http://ollama-cpu:11434
       - API_SERVER_ENABLED=true
       - API_SERVER_PORT=8642
       - API_SERVER_HOST=0.0.0.0
@@ -96,12 +96,12 @@ services:
       - "traefik.http.routers.syncthing-https.tls.certresolver=njalla"
       - "traefik.http.services.syncthing.loadbalancer.server.port=8384"
 
-  ollama:
+  ollama-cpu:
     build:
       context: ./ollama
       dockerfile: Dockerfile
     image: ollama/ollama:rocm-gfx906
-    container_name: ollama
+    container_name: ollama-cpu
     tty: true
     restart: always 
     ports:
@@ -114,14 +114,6 @@ services:
       - OLLAMA_VULKAN=0
       - OLLAMA_HOST=0.0.0.0
 
-networks:
-  ai_net:
-    external: true
-    name: ai_net
-  ai_backend:
-    driver: bridge
-    name: ai_backend
-    
   llama-cpp-hermes:
     image: llama-cpp:rocm-gfx906
     container_name: llama-cpp-hermes
@@ -157,6 +149,14 @@ networks:
       --no-mmap
       --n-predict -1
 
+networks:
+  ai_net:
+    external: true
+    name: ai_net
+  ai_backend:
+    driver: bridge
+    name: ai_backend
+    
   # vllm:
   #   image: nalanzeyu/vllm-gfx906:v0.9.0-rocm6.3
   #   container_name: vllm