Say I call a chat completion as follows:

 OpenAIPromptExecutionSettings settings = new()
 {
     ServiceId = "swedencentral",
     FunctionChoiceBehavior = FunctionChoiceBehavior.Auto()
 };
 kernel.Data.Add("service_id", "swedencentral");
 var reply = await kernel.InvokePromptAsync("Some question?", new(settings));

If I register an IAutoFunctionInvocationFilter I cannot get access to the prompt execution settings to get the service id.

The workaround is to use kernel.Data but this is not ideal.

The problem is in the FunctionCallProcessor as it does not pass on the original execution settings.