Update custom llm model url response

[pkulkark]

Description

This PR updates the custom llm models url response handler so that it can work with responses with the format like this:

{
  "models": {
    "Meta-Llama4-Maverick": {
      "name": "Meta-Llama4-Maverick",
      "display_name": "Maverick (Llama 4)"
  }
}

Testing instructions

1. Deploy this branch of the XBlock.
2. Enable custom service in Site Configuration. In your Open edX Django admin → Site Configurations, add:

{
    "ai_eval": {
             "USE_CUSTOM_LLM_SERVICE": true,
             "CUSTOM_LLM_MODELS_URL": "https://your-custom-service/models",
             "CUSTOM_LLM_COMPLETIONS_URL": "https://your-custom-service/completions",
             "CUSTOM_LLM_TOKEN_URL": "https://your-custom-service/oauth/token"
    }
}

3. Set the following client credentials in Django settings for Oauth2 (e.g. via Tutor’s configuration):

CUSTOM_LLM_CLIENT_ID = "your-client-id"
CUSTOM_LLM_CLIENT_SECRET = "your-client-secret"

4. Add the XBlock component and verify that the models dropdown is populated correctly with the available models list.

Other information

Note: display_name is currently ignored because the dropdown expects a list of model strings which are sent in the request.

[carlos-marquez-wgu]

Hi @pkulkark , we tested this in our side is working, thanks!