fix(oci): label Ollama API endpoint accurately

oci/catalog/overlays/ollama.json

@@ -1,4 +1,23 @@
 {
+  "catalog_ui": {
+    "description": {
+      "en_US": "Ollama runs large language models locally through a command-line interface and an HTTP API. The Ollama server does not include an interactive Web UI; pair it with a frontend such as Open WebUI if you want browser-based chat.\n\nIt supports many models, including Llama, Mistral, Gemma, and Qwen, and provides an API for applications and integrations. Local models run on your own hardware; cloud models require an internet connection.\n\n**Key Features:**\n- **Local model execution**: Run supported models on your own hardware\n- **Model management**: Pull, run, and manage models with the CLI or API\n- **API integration**: Connect other applications to the Ollama server\n- **Privacy options**: Keep inference local when using local models\n\n**Learn More:**\n- [Ollama Official Website](https://ollama.com/)\n- [Ollama GitHub Repository](https://github.com/ollama/ollama)\n- [Model Library](https://ollama.com/library)\n"
+    },
+    "launch": {
+      "path": "/api/version"
+    }
+  },
+  "first_run": {
+    "endpoints": [
+      {
+        "label": "Ollama API (no Web UI)",
+        "scheme": "http",
+        "port": 11434,
+        "path": "/api/version",
+        "source": "ollama-http-api"
+      }
+    ]
+  },
   "proxmox": {
     "deployment_profile": {
       "variant": "hardware-selectable",