bug#16733: messed up unicode chars in package description

[Eli Zaretskii]

> From: Juanma Barranquero <lekktu@gmail.com>
> Date: Thu, 20 Mar 2014 06:12:24 +0100
> Cc: Eli Zaretskii <eliz@gnu.org>, 16733@debbugs.gnu.org
> 
> 1) Leave it as it is now, with
> detect-coding-string/decode-coding-string (or perhaps
> decode-coding-(inserted-)region).

This is suboptimal when the URL already tells its charset.  In that
case, you should use the information; you should fall back to
detect-coding-string only if that information is not available.