How does inferencing support streaming responses?

Centralpoint supports server-sent-event streaming from cloud LLMs and from local models, so end users see tokens as they are generated rather than waiting for the full completion. This matches the conversational UX users expect from consumer chat tools. The plat