Add straming_callback to run() method of OllamaGenerator #315
Description
Currently, you can only set the streaming_callback on OllamaGenerator through the init method, which sets it for the entire object. This is problematic if you aim to re-use this object to handle concurrent requests, because the streaming callback typically references different chunk collectors (for different requests).
As such one cannot have concurrent processing of requests by the same pipeline.
Suggested implementation:
Just add streaming_callback argument to the run() method, so that it can be dynamically set with each request. (like OpenAIGenerator for instance)