.Net: Added support for multiple chat and text results from Kernel

[dmytrostruk]

Motivation and Context

Note: This PR changes the behavior when AI connector returns multiple results, when using kernel.InvokeAsync. It won't throw an exception anymore, but instead it will return multiple results. The behavior for single result is not changed.

Fixes: #6434

When executing prompt function using kernel and asking for multiple results per prompt, we will get an error:

var arguments = new KernelArguments(new OpenAIPromptExecutionSettings
{
    ResultsPerPrompt = 3
});

var result = await this._kernel.InvokePromptAsync("Hi, can you help me today?", arguments); // this will throw an exception

Current KernelFunctionFromPrompt implementation expects only single result from AI connector, while its API can return multiple results per prompt/request.

This PR updates KernelFunctionFromPrompt in a following way:

1. If AI connector returns single item - the behavior will be the same as it is today, FunctionResult will contain instance of that item, so it's possible to get its properties, use ToString() etc.
2. If AI connector returns multiple items - all items will be returned in collection to the caller, and this collection needs to be handled appropriately (by using loop or accessing specific item by index).

One of the examples shows how to select one result, in case if we invoke prompt function inside another prompt function using prompt template engine. In this case, filter can be registered, which will get multiple results produced by function, select one of them and return it back to the prompt rendering operation.

Contribution Checklist

• The code builds clean without any errors or warnings
• The PR follows the SK Contribution Guidelines and the pre-submission formatting script raises no violations
• All unit tests pass, and I have added new tests where possible
• I didn't break anyone 😄