Charset cleanup

[Lars Magne Ingebrigtsen <larsi@gnus.org>]

I'm going to do a re-design of the charset variable thingies.  Here's
my ideas for now:

1) All the mm-*charset variables will default to nil.

2) The same with all rfc*charset variables.

This means that the default for these packages will be to encode using 
the MIME methods.

3) There will be a gnus-default-charset variable which says what the
"locale" is to be for unknown groups.  This will default to nil, which 
means that all messages to all unknown groups will be encoded.

4) There will be a gnus-group-post-charset-alist variable, which says what
the default charset in groups that match the keys will be.  The
default will include things like

  ("^no\\." iso-8859-1)

which means that characters from this charset wont be RFC2047-encoded
when posting to Norwegian groups, for instance.  I need a list of
hierarchies that have similar rules, and what charsets they default
to.  Mail will default to full MIME encoding.

5) There will be a gnus-group-read-charset-alist variable, which
specifies what charset Gnus is to assume articles with no charset
specification is supposed to be in.  This will overlap greatly with
the previous variable, but this latter variable is more "sloppy".

6) Gnus or Message won't ever set any of the rfc*charset or mm*charset 
variables, but may bind them, or set them locally in buffers.

Zhu, I realize this will kinda collide with many of your nice fixes,
but I just think that there's too many variables around now.  Which is 
my fault.

I'll release 0.70 first (to get rid of the infloop thing), and then
get right to it.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen