Improve parsing of multipart (HTML) emails

When JIRA is configured to create/comment issues from email, it may encounter an email with arbitrary MIME parts (text, HTML, images, videos, etc). Currently JIRA will try to rip such mails apart, attempting to extract text from the HTML, and attaching other bits as attachments.

This is a problem because:

• No attempt is made to identify "alternative" parts which mean the same thing. In the case where there is a text/plain part and a text/html part, JIRA will extract the text/plain part as the body of the message, but then as well attempt to attach the text/html part to the issue. This usually fails however, as text/html parts when inline in a message usually denote the HTML version of the plain text, and thus mail clients don't usually bother to attach a filename to the part. This causes a warning message to appear in the logs (see JRA-10825 for more info):

  Cannot create attachment without a filename - inline content?

  We need to come up with a way to identify these alternative parts, and ignore the redundant parts so that they are not attached to the JIRA issue.

• This may completely lose the meaning of the email. For example, the position of inline images may have an impact on the meaning of the message.