[kbenson]

If by "rerouting" you mean user level forwards, then that's generally done at the mailbox level, but not the client side (depending on system and rules), because users are allowed to set their own forwards.

Mail loops in mail servers have always been a thing, and there are various ways to detect and stop them, usually by adding headers to a message and detecting it's the same one that was seen previously.

If there's actually a client side forward (which mail servers can't prevent because running mail through a local program, whether it be procmail, spamassassin, or some other mail categorization and filtering program) is a valid use case with a long history. There's also the case where the mail client is applying it's own complex rules and sending responses automatically. If the message is actually a new one, there's not an easy way to detect the loop.

There's a reason why Gmail for a long time (still?) didn't allow you to forward your mail to another Gmail address. I would argue that for Exchange, where it's generally much easier to track down individuals, erroring on the side of allowing administrators to do what they want to block this (whether that be company policy, exchange policy, or whatever) is the correct design choice.