Visualize "style" of LLM responses across particular questions

Right now, users have to glance at responses manually in an `InspectNode` to gleam how LLMs differ. This becomes difficult especially for large quantities of responses where num responses per prompt `n > 1`. Ideally, there should be a way to visualize the "style" of a particular LLM response (for particular sets of input prompts and prompt parameters). This involves designing, or adopting, a kind of aggregate text visualization and aggregation technique that presents the 'style' of responses per set.