fix: support overrideName for Responses API deployments

[adubovik]

Before the fix: the following DIAL Core configuration for Response API deployment wasn't possible:

{
    "models": {
        "my-dial-deployment": {
            "type": "chat",
            "overrideName": "gpt-4o-2024-11-20",
            "endpoint": "${ADAPTER_OPENAI_ORIGIN}/openai/deployments/my-adapter-deployment/chat/completions",
            "upstreams": [
                {
                    "endpoint": "${AZURE_OPENAI_ORIGIN}/openai/v1/responses",
                    "key": "api-key"
                }
            ]
        }
    }
}

Adapter tried to call Azure OpenAI Response API with a model name equal to my-adapter-deployment.
This request was failing since such an Azure OpenAI deployment doesn't exist.

After the fix: the adapter correctly calls the model called gpt-4o-2024-11-20 provided in the overrideName parameter. The call succeeds (assuming the deployment indeed exists).

---

Added unit test for this.

[adubovik]

/deploy-review

Environment URL: review-environment | pipeline
Chat E2E tests status: success
Overlay E2E tests status: success