Definitely not appropriate on the following grounds- all obvious.
* the instuctor can not monitor the students progress correctly and watch for the subtle inconsistancies in critical areas like TO because alcohol lowers inhibition a mindset of " to heck with it let em go below DH without a visual, I don't care" may ensue
* it is a bad message
* a simulated event should be as real as possible a noticeably drunk instructor would dampen the "reality play" required as such events during training are highly unusual
as devil's advocate--- a few dollars wasted on such a guy would prove a valuable CRM lesson ?-I guess.
there's probably a billion reaons against, and a million for.