Add format support for Ollama generators

[marfago]

Is your feature request related to a problem? Please describe.
Ollama has introduced structured output as reported here.
This feature now allows developers to define a json schema which represents the expected output format

Describe the solution you'd like
Adding the support should be straightforward since it just require to add and propagate a new parameter 'format' to the generator. It may look like:

generator = OllamaGenerator(
    model="llama3.2",
    "format":aPydanticModel.model_json_schema()
)

Describe alternatives you've considered
N/A

Additional context
The Ollama launch blog

[julian-risch]

Related issue: deepset-ai/haystack#8276

[SotirisKot]

A simple hack for now is to override/extend the OllamaGenerator component. Example below:

# override the current OllamaGenerator to account for structured outputs
# since haystack does not currently support this
@component
class StructuredOllamaGenerator(OllamaGenerator):
    def __init__(self,
            model: str = "orca-mini",
            url: str = "http://localhost:11434",
            generation_kwargs: Optional[Dict[str, Any]] = None,
            system_prompt: Optional[str] = None,
            template: Optional[str] = None,
            raw: bool = False,
            timeout: int = 120,
            keep_alive: Optional[Union[float, str]] = None,
            streaming_callback: Optional[Callable[[StreamingChunk], None]] = None,
            format: Optional[Dict[str, Any]] = None
        ):
        super(StructuredOllamaGenerator, self).__init__(
            model=model,
            url=url,
            generation_kwargs=generation_kwargs,
            system_prompt=system_prompt,
            template=template,
            raw=raw,
            timeout=timeout,
            keep_alive=keep_alive,
            streaming_callback=streaming_callback
        )
        self.format = format
    
    @component.output_types(replies=List[str], meta=List[Dict[str, Any]])
    def run(
        self,
        prompt: str,
        generation_kwargs: Optional[Dict[str, Any]] = None,
    ):
        """
        Runs an Ollama Model on the given prompt.

        :param prompt:
            The prompt to generate a response for.
        :param generation_kwargs:
            Optional arguments to pass to the Ollama generation endpoint, such as temperature,
            top_p, and others. See the available arguments in
            [Ollama docs](https://github.com/jmorganca/ollama/blob/main/docs/modelfile.md#valid-parameters-and-values).
        :returns: A dictionary with the following keys:
            - `replies`: The responses from the model
            - `meta`: The metadata collected during the run
        """
        generation_kwargs = {**self.generation_kwargs, **(generation_kwargs or {})}

        stream = self.streaming_callback is not None

        response = self._client.generate(
            model=self.model, 
            prompt=prompt, 
            stream=stream, 
            keep_alive=self.keep_alive, 
            options=generation_kwargs,
            format=self.format
        )

        if stream:
            chunks: List[StreamingChunk] = self._handle_streaming_response(response)
            return self._convert_to_streaming_response(chunks)

        return self._convert_to_response(response)

You must have the latest ollama version of course.