.Net: Extend the ModelResult interface to track Max token exceeds

Currently, it's cumbersome to track when the creation is interrupted due to exceeding the max tokens. In C#, it would be helpful to include a finish_reason in the ChatModelResult so that developers can handle it when the max_token is exceeded.

// semantic-kernel/dotnet/src/Connectors/Connectors.AI.OpenAI/AzureSdk/ChatModelResult.cs
    internal ChatModelResult(ChatCompletions completionsData, ChatChoice choiceData)
    {
        this.Id = completionsData.Id;
        this.Created = completionsData.Created;
        this.PromptFilterResults = completionsData.PromptFilterResults;
        this.Choice = choiceData;
        this.Usage = completionsData.Usage;
        // Add this.FinishReason = completionsData.FinishReason
    }

While Python currently only returns a simple string, I hope it to be changed to return an ChatModelResult object, similar to C#. (If these need to be opened as separate issues, I will split them accordingly)

I am using the Python version and indirectly controlling it through the following method:

def is_output_length_exceeding_max_tokens(plugin: SKFunctionBase, output_str: str):
    try:
        encoder = tiktoken.encoding_for_model(plugin._ai_service._model_id)
    except KeyError:
        encoder = tiktoken.encoding_for_model("gpt-3.5-turbo")

    num_tokens = len(encoder.encode(output_str))
    return num_tokens >= plugin.request_settings.max_tokens

microsoft / semantic-kernel

.Net: Extend the ModelResult interface to track Max token exceeds #3119