Some data points you can use to identify the sender’s email client are the MIME boundary delimiter, Message-ID, Content-IDs, and quote strings. Quote strings can sometimes help identify the email clients that were used to compose previous messages in the email thread as well.
As you mentioned, trace headers may also contain clues. For instance, they may contain references to the transport that was used such as MAPI, which would help narrow things down.
We had a relevant discussion here:
For reference, here is what FEI does with that information on a simple sample message. This message did not have transport or Content-ID evidence, but you can see how the Message-ID, MIME boundary, and quote string evidence are used toward client identification.
