Consider nupunkt for case name detection

[jcushman]

This repo and research paper sounds relevant for tokenizing sentences, and therefore being able to properly detect case names: https://github.com/alea-institute/NUPunkt

[mlissner]

@flooie, @rachlllg took a look at this today for use in semantic search but found it wasn't a good fit there, but Jack's right that it might make sense in eyecite. Want to take a look to see if it helps with case name identification?

[rachlllg]

Just want to chime in with some insights here. Michael recommended considering using nupunkt in conjunction with charboundary for more balanced performance, since nupunkt prioritizes precision over recall, also if using charboundary, be sure to use the onnx version for speed boost. charboundary is a lot slower than nupunkt but we can use parallarization to combat some of that.

https://arxiv.org/abs/2504.04131