join -> concat, write->writeln, make errors consistent

Created by: LazyOxen

Think I got everything except for the parsing numbers in the tokenizer. I'm leaving that alone for now because I'm thinking about making some other changes that pertain to that.

Also, I think input.len() is a bit generous for allocating tokens...essentially each character gets a token, which feels a bit wasteful to me. I suppose in practice it might not make much of a difference since most lines probably won't be a million characters long.