Unicode strings.  At least MIME::Parser::parse_data() and
Template::proccess(): these methods occationalily break
strings containing Unicode characters.

My mail_utf8 patch expects the behavior as following ---

Sub-messages to be attached (into digests, moderation
notices etc.) will passed to mail::reformat_message()
separately then attached to reformatted parent message
again.  As a result, sub-messages won't be broken.  Since
they won't cause mixture of Unicode string (parent message
generated by tt2::parse_tt2()) and byte string (sub-messages).

Note: For compatibility with old style, data passed to
mail::reformat_message() already includes sub-message(s).  Then:
- When a part has an `X-Sympa-Attach:' header field for
internal use, new style, mail::reformat_message() attaches
raw sub-message to reformatted parent message again;
- When a part doesn't have any `X-Sympa-Attach:' header
fields, sub-messages generated by [% INSERT %] directive(s)
in the template will be used.
See src/mail.pm hunk #4.

More Note: Latter behavior above will give expected result
only if contents of sub-messages are US-ASCII or ISO-8859-1.
In other cases customized templates (if any) should be
modified so that they have appropriate `X-Sympa-Attach:'
header fileds.

Sub-messages are gathered from template context paramenters.
  See src/mail.pm hunk #1.

Other trivial fixes:
  mail_tt2/listmaster_notification.tt2 hunk #1
  mail_tt2/request_auth.tt2
  src/Commands.pm: informations of list in INFO message will
be translated.
  src/mail.pm hunk #5: fallback to UTF-8 when locale of user
was not same as that of list.


git-svn-id: https://subversion.renater.fr/sympa/trunk@4007 05aa8bb8-cd2b-0410-b1d7-8918dfa770ce