[Evolution] 8 bit header changes

[NotZed <notzed ximian com>]

Hi guys,

I've changed slightly the way evolution handles headers, like subject, etc.

It'll try to convert headers that come in, using the charset defined
for the main body of the message's content-type field, and if that
fails, then use the charset defined for the current locale, and then
if that fails, assume its iso-latin1 or ascii.

This should work ok for most cases:
 - properly encoded 7 bit-clean headers
 - mail sent by a client that uses the mail body 
 - mail sent by a client that uses the locale charset

It wont work in some cases that represent valid mail:
 - Where the subject contains data in ascii format, but the header
   charset or locale charset doesn't overlap ascii.

I can fairly easily fix the last problem by only converting data
that contains 8 bit characters, otherwise assume it is ascii, but
i'm not sure if that would be more inconvenient for some users
(japanese?  korean?); the reasoning being that native language
not displaying right is more a problem than english not displaying
right.  Of course, if its basically impossible to represent real words
in such languages in 7 bits of non-overlapping ascii charset, then it
should only convert if the data is 8 bit.

Let me know if this causes any weird problems.

Its only just been checked in, but should be in the next snapshot.

 Michael