That's not really a solution, that's a work around to a problem that shouldn't exist in the first place. The number of guests is far too high and looks unrealistic at base level, and that number is causing people's performance to tank and forcing people to play the game differently when they shouldn't have to. There's no benefit to keeping it the way it is. If zoos are frequently becoming too crowded to be playable, then the game developers should reduce the crowding and balance the ramifications around the decision for the sake of the fun. Your entire argument seems to be that AI is too resource intensive, my solution is to reduce the number of AI running in the park significantly and rebalance other elements of the game around that. If you have 1000 individual guest AIs running, every single decision that a guest needs to make is running 1000 times; if you have 200 guests, those same lines of code are only running 200 times. The performance savings in down scaling number of guests would be well worth it and allow resources to be invested elsewhere.
Edit: I wanted to add some numerical proof. So the San Diego Zoo, one of the largest and most popular in the US, hit 4 million visitors annually for the first time in 2018. They're open 365 days a year, so that averages out to roughly 10959 guests per day. They house more than 650 species and subspecies, and over 3500 individual animals. None of our zoos are that scope. None of our zoos are half or a quarter or even an eighth of that scope (an eighth would be 81.25 species, so we're actually pretty close!). So why are our visitor numbers? If we assume that number of zoo guests scales linearly with the number of species displayed, an amazing zoo with everything in the game currently really shouldn't have more than 1500 visitors at a time.