To answer the very question asked,
why would we not just use 는/은 for that (그 것은 알고 싶다 , 사과는 먹고 싶어요)?
that's because it would sound very strange.
I can assure you that no native speaker would say
그것은 알고 싶다
in isolation. He might say:
그것은 알고 싶고, 이것은 알고 싶지 않다.
Here the contrast justifies 은 and 는.
I find your hypothesis of 그것이 or 사과가 acting as de facto subject quite insightful.
Grammar is after all only a theory. The aim is to reduce all phenomena (here, usages) to a decent set of basic concepts. Because the number of such concepts should not be too great, we sometimes have to make awkward choices. The choice could go either way. That is to say, internally consistent grammar would be possible in which 그것이 or 사과가 is either subject or object.
The awkwardness comes out well if you consider this (different) example:
나는 배가 아프다.
If you say that 나는 is subject and 배가 a complement, then it preserves the principle that a sentence should have only one subject. This however comes at the cost of having to say that, in
배가 is subject. Conversely if you want consistency of treatment for 배가, you may have to allow a sentence having two subjects sometimes. No doubt some other treatment is also possible.
You may come to a quandary like this in any language I believe. For example, 'convince me that P' sounds rather like 'give me that book'; and it would be internally consistent grammar to say that 'me' was indirect object in both cases (or, if you will, dative). But some people will get vehement about what they learned in Mrs. Gault's honors English back when.
Further on your hypothesis, consider:
This car rides well.
There are two things we can say.
'Car' is the subject. But 'ride' has a different meaning than in 'I ride alone.' It means something like 'is ridden well.'
'Car' is the object in spite of heading the sentence. The sentence is simply 'Null-object rides this car well.' We use this construction when we want to say that anyone would find that he could well ride a particular car.
(Those who wish, may now get vehement about 1 being right and 2 wrong, or perhaps both being wrong.)
What Korean grammar has done for,
사과가 먹고 싶다.
is analogous to 2. That is, it has paid the price of assigning multiple syntactic roles to 이/가 so as to keep the semantics of 먹고 싶다 simple.
But there is nothing to prevent you from doing 1. I.e. to say that 사과가 is the subject, but that 먹고 싶다 has a different meaning, something like that of 'is desired for eating'--if that helps comprehension. (But obviously not on an exam.)