you have to wonder... what is a greater threat to the bottom line?
the amount of resources and time needed to create this "feature"
OR
selling devices with an adequate amount of storage in 2015
I'll bite.
Resources and time needed for this feature: 2 programmers for a few months, 5 testers for a few weeks, a few dozen marketing type folks for localization and documentation purposes. Rough estimate: one-time cost of $ 200k, taking > 100% contingency I think you can safely say the cost is max $ 500k. In fact the biggest cost is probably the opportunity cost of these people NOT focusing on something else.
Selling 100 million iPhones per year with 2$ cheaper memory inside: recurring benefit of $ 200m. Now of course people won't pay the same price for a 32 GB iPhone then they do for a 64 GB iPhone so that doesn't quite hold true. But people are implicitly paying for usability, not for memory size by itself. If you combine a boatload of tricks like "App Thinning" and "On-Demand Resources" and this "Temporary Delete to free up Space" to squeeze 64 GB of usability out of 32 GB of hardware - you can sell a smaller phone for a higher price. So let's assume on the safe side they can get 10% of that benefit - that's still a 20m $ return on a 500k $ investment.
There's also the notion that this feature probably fits into the bigger picture of what they're trying to achieve with things like On Demand Resources, where the OS proactively manages the storage on the device to proactively delete app assets to make room for other assets. I wouldn't be surprised if the idea of iOS proactively deleting the least used apps and then re-installing it as required will appear as an option in iOS 10 not just for iOS updates but for any app installations, and what they're doing here is basically testing the water to see how people respond to such a notion and start getting them used to it.
Note these ideas of local storage optimization keep applying on all future generations of these devices. Since memory and app sizes can be expected to scale up together this will perpetually remain a problem and this feature is a perpetual solution - replacing the 16GB model by 32GB is a one-time fix with no future benefit. Network speeds have historically been increasing faster then storage costs have been decreasing; extrapolating this 5-10 years it should be no surprise if the vast majority of your data is simply downloaded as required and local storage is only a cache. This probably seems a bit unnatural given today's network speeds but given technology trends it seems inevitable. This is probably the future that Apple is building for (as the prophet used to say: "skate where the puck is going, not to where it's been).