Add adapter classes for Microsoft.Extensions.AI

[Nukepayload2]

As an app developer, I need to let end users control where the LLM model should be running. For Llama models running on local computer in the application's process, we use LlamaSharp. If the user want to use an AI service on the Intranet, we use Ollama. If LlamaSharp implements https://www.nuget.org/packages/Microsoft.Extensions.AI.Abstractions , the model provider logics would be much easier to implement.

[martindevans]

Stephen Toub added support for these classes in #964, does that work for you or are there some bits missing?

[Nukepayload2]

Sorry, I mistakenly thought that the Microsoft.Extensions.AI integration doesn't exist, because I didn't find anything like ILLamaExecutor.AsChatClient() or Microsoft.Extensions.AI in the documentation.