New parsing in 0.15

[Boris 'pi' Piwinger]

Hi!

My first tests with 0.15.0 are successfull. I hope the new
method generates a more robust system (so all tricks to
avoid running into the next message can go).

Could and should the following be done:

1) Unfold header lines (i.e., replace newline followed by
whitespace with that whitespace).

2) Remove whitespace between adjacent MIME words.

If this would be done before the messages is analyzed, we
would get rid of all those problems with header fields we
have now.

pi