journal archive email format and dlp issue

I think it's the envelope journaling that is the key here.  The message itself  (P2) is actually submitted using the sender/recipient cannonical name, and the hub transport server will add the SMTP address as it puts the message in its envelope (P1).  See MS for more detail on how envelope journaling works with internal addresses: http://msdn.microsoft.com/en-us/library/office/cc842372.aspx

  When EV stores the item in the archive, the P2 is stored since it is the actual message, and the information in the P1 is written to the index and  saveset metadata, and the P1 envelope is discarded.  When you perform a PST export, you are not recreating this envelope. You are just exporting the actual message, and the SMTP address is not written on the actual message.

FWIW if you enable the API then DA gets the sender/recipient address by going through the API to read the saveset metadata  but I believe that is just for review and not for exports.