Refactor: need clarification for TODO (shared base for Chat and Completion)

[greninja]

I just need a clarification as I am planning to open a PR for this. There’s a TODO in the codebase about unifying “Chats and Completions” (rigging/completion.py L34–35)

The TODO doesn’t say whether that means the Chat and Completion classes (what you get back from .run()) or the ChatPipeline and CompletionPipeline classes (what you call .run() on). Which pair should we unify under a shared base (Chat and Completion, or ChatPipeline and CompletionPipeline)

There’s duplication in both places: the Chat/Completion models share fields like uuid, timestamp, metadata, stop_reason, usage etc. The pipeline classes share generator, params, metadata, and methods like with_(), catch(), and clone().

I’m happy to implement this in a follow-up PR