I want to combine the available CoreML models that Apple provides in order to create a large one that is able to identify objects more accurately. Is there a way to either unpack these models, combine them, and repack them, or to use them in an array combining within the Swift code?
In short: no. This is not how these models work.
You could combine multiple models into one big one, but that's the same as running them individually.
What you can do is make predictions using multiple models, then take the average of these predictions, and use that as the final prediction. That's called "ensembling". But it's overkill for use on a mobile device.