Had same problem (including down to following Mitch's innocent script).

There is a ton of cache going on between client machine and the DNN instance, much outside your (at least my...) control.  Until I took down the old instance/path completely (and removed the new alias from the old path/DNN), 'requests' seemed to still get pointed to and served by the old instance.  [I couldn't wait 24-48 hours for all the routers in the world to get caught up...]

If you wind up in a seat on the wrong train but with a ticket that looks right, you continue to get served by DNN; it's the whole point of the alias approach...

DNN doesn't care how you get there, if you land in a DNN directory with a ticket (alias) for a valid portal, your request gets served.  Even though it is not what you want any more...

You did clear client cache/history?