OpenAI tests are hardcoding model names rather than using the specified model

[ericstj]

This is breaking testing on Azure OpenAI when the hardcoded model is not deployed.

https://github.com/search?q=repo%3Adotnet%2Fextensions+gpt-4o-mini+language%3AC%23+path%3Atest%2FLibraries&type=code

extensions/test/Libraries/Microsoft.Extensions.AI.OpenAI.Tests/OpenAIChatClientTests.cs

Line 151 in 20db541

using IChatClient client = CreateChatClient(httpClient, "gpt-4o-mini");

extensions/test/Libraries/Microsoft.Extensions.AI.OpenAI.Tests/OpenAIResponseClientTests.cs

Line 149 in 20db541

using IChatClient client = CreateResponseClient(httpClient, "gpt-4o-mini");

Etc