Must we reject intro_established extensions earlier?

See comment in ipt_establish.rs:

 // TODO HSS this check needs to happen in IptMsgHandler::handle_msg,
 // because otherwise handle_msg might go on to handle messages despite
 // us wanting to crash, here.  (Providing reliable teardown of the
 // IptMsgHandler wouldn't be sufficient, since there would be a race.)

We should either resolve this issue, or document why it doesn't need to be resolved.