Yeah, I wasn't sure at first, but on second read, it seems that "questionable" in this context means "may break when piped through a system that thinks in latin-1."
This could be useful, then, for finding curly quotes that make their way into templates or HTML files (often by way of someone pasting text from e.g. Microsoft Word) and then subsequently break old, non-unicode template-parsing utilities.
Questionable is probably not the right word. The idea is to find characters that you might be surprised are in your document. It's specifically helpful for zero width or invisible characters, which is replaces with their name.
Future revisions might say "Most of your characters are European but you have this one Kanji character, let's highlight that"
Try "hello" in Polish: cześć