Tasks

[danielmlow]

Description

High priority

• Add examples of the output DF (LLM coding and summary of metrics for a few constructs) in the README

Low priority

• Return reasoning on top of the verbatim quote: more advanced models (Gemini Pro I think) automatically return reasoning for the score in the metadata. Some advanced models also provide a more high-level reasoning in the metadata, so we should return that as well, but this latter one won't be the reasoning for each construct which is what we want.