String-based line-wrapping for formatter

[billy-bao]

This is a basic implementation for line-wrapping LF code. The formatter takes into account the current level of indentation when breaking lines for line-wrapping, but will always break at spaces (violating the line-wrap limit if it cannot find a space before the limit). Set the line-wrap limit with ToLf.instance.setLineWrap().
Part of #1227.

[petervdonovan]

This implementation is way simpler and less messy than what I had in mind.

I left some nitpicks. My only real concern with this approach is what I said on Slack -- that to do this at the level of string processing as opposed to AST processing, while respecting single-line comments and strings and Python code blocks etc., will effectively result in a re-implementation of part of our parser. We have done something like that with semantic highlighting, but it was a little messy and I am not sure if it was a good idea (as in, we might eventually delete those 500 lines of code and replace them with queries to the language server, which has the AST).

I advocate for merging this into the pretty-printer branch now so that we can move on to other things, with the understanding that if we do run into the problems that I have talked about, I am not in favor of investing a lot of time into getting this to work well without directly processing the AST.