Use correct char set for command-line scripts

The command-line scripts in mailman-2.1.5 share the i18n
framework with the rest of mailman, but unlike the rest
of mailman,
mailman is not in control of the characer set used; it
should always
use the character set specified by the LC_CTYPE locale
category.
Currently on a system using e.g. LANG=cs_CZ.UTF-8 the
Czech (cs) translation is selected, but the output uses the
character set of the translation (ISO 8859-2), not UTF-8.

The attached patch (roughly against 2.1.5) adds
Mailman.i18n.C_(),
which works like _(), except that the output is in the
LC_CTYPE-specified encoding, and modifies all command-line
utilities to use it.