Some languages such as C# default to a UTF-8 charset for source code. The default ArcanistTextLinter prevents all non-ASCII character in a text file, so this adds an option which can be used to turn off bad charset detection.
Ideally there'd be an option to turn on UTF-8 source code mode instead of just a blanket ignore (since then it could also detect when the source code will trip up arc diff's unicode detection), but I don't know enough about character sets to implement UTF-8 validation.