I'm fairly certain that rfc2821 allows the initial mta to remove the return-path header if it exists, so it is not incorrect to send the entire message file when re-injecting the message.
Allowing an MTA to remove a header that should never have been sent is not the same as allowing the header to be sent in the first place.
So, no, it is not correct to include the Return-Path header when forwarding or bouncing the message.
-Pat
