generalize line ranges

[wrongu]

option 1: move to char ranges

option 2: move to 'diff' semantics rather than selection ranges

...

option N: see discussion below

[j6k4m8]

https://github.com/KordingLab/llm4papers/pull/17/files#diff-be91071fd78206cad9b9beab69c69678187bbcdc69f97d65d2dd9563cd3fd2afR42-R75

[wrongu]

on the one hand, would be nice for there to be a "document location" protocol, but then we can't put them into pydantic models.

still need to think more about the "right" way to have one data structure that can speak the languages of character offsets, parsed document trees, line ranges, etc. And think about how far we can get with quick and dirty hacks (like pretending everything is a character range)