[Groff] preconv now in groff CVS

[Werner LEMBERG]

I've just committed a first version of preconv, a preprocessor for
groff which converts various input encodings to something groff can
understand.  Still missing is integration within the groff binary and
autoconf stuff for proper handling of iconv and locale functions.

Almost nothing has remained from Tomohiro's first draft.  Main reasons
for the changes are as follows.

  . Updated emacs-to-mime list to current CVS version of Emacs.  There
    are still some strange entries, either from older Emacs versions
    or from XEmacs.  Any help to fix this is highly welcome.

  . Removed mime-to-locale functions which are not necessary IMHO.

  . Added UTF-8 parser.  This code should eventually go to GNU troff.

  . BOM (Byte Order Mark) handling added (this is part of the UTF-16
    and UTF-32 specification).

  . Encoding tag recogition now works even with UTF-16 and friends
    (something which Emacs doesn't support yet -- I've already files a
    bug report).

  . Encoding tags are now searched in the first two lines only.

  . A `\#' comment (starting a line) is supported also.

  . Output of \u[XXXX] entities.

  . cp1047 handling uses hex values only.

  . Input data is no longer completely read.  This, together with the
    BOM stuff took a lot of time to make it work.

  . iconv handling completely rewritten to use small input and output
    buffers.

Please test!


    Werner