If the key is only good to decrypt a random part of the movie, then the test says nothing about the success or failure of the key to decrypt the rest of the movie.
Tangentially related: what is to stop a technically inclined projectionist to use the key during the time it is valid to decrypt the movie?
I would assume that at presentation time, there is enough of decryption key material available to the projection system to cover all of the movie, otherwise a presentation would have to be interrupted as the cinemas internet goes down.
> One of those would be manna from heaven for any unsavory projectionist who wanted to make high quality copies from theater prints.
Not really. The keys and image decryption happen in a tamper-resistant "FIPS can" in the media server and then are re-encrypted and sent across a bus to the projector. Some newer projectors have the FIPS can directly inside so that there are no physically exposed interfaces with anything other than the full-blown AES encrypted frames. Having a KDM means you can play the movie but you can't ever actually get to the data.
Fun fact: before DCinema there used to be people who owned tele-cine scanners mounted in the back of flatbed trucks that would go around to theaters and pay unscrupulous employees to borrow the film for a couple hours. They'd have an very nice scan of the movie for pirate DVDs the day the movie came out.
This seems like a simple and critical feature. If the distributor wishes to impose these extreme constraints on the projectionist, then the projectionist needs to know its going to work well in advance. The projectionist shouldn't even need to be in the room when the movie starts.
I've run into this pattern a lot in web development. If you want a feature to go live at 6:00pm, you don't release it at 5:59pm and just hope it works. You program it to go automatically flip at 6:00pm and release the code early. The pre-programmed time can be overridden in the production environment, so you can easily test and know for sure (or as close as possible to sure) whether it will flip at 6:00pm or not.
(I'm not saying my solution is perfect, just better than the naive solution.)