Improve handling of Outlook emails which contain html, and those with non-unicode text bodies

There are two areas in the outlook parsing that could use some enhancement:

• If the text content is stored as non unicode strings, we should try to have the right encoding used for this
• If the email contains a html or rtf version of the message body, we should prefer this over the plain text one

Note - some of these enhancements may need to wait for POI 3.8 beta 3