Fixing a regression introduced in #288. We decided that Unicode characters like 　 should be copied to the clipboard as-is and only be escaped for displaying them, but the way I made this change also resulted in the ASCII characters \n \t \r and \ being copied as-is too, which is annoying when you want to paste these strings into source code. So this reverts ASCII characters to their previous behavior, but still leaves Unicode characters like 　 unescaped.