But it does stop being science when you abandon testable predictions.
No it doesn't. This is a common myth, but if you look at astronomy or biology you will obviously see that is not the case. Observational science does not require experiments.
Yes, experiments provide more rigorous evidence, but to deny that observation also provides evidence is quite frankly to throw the baby out with the bathwater. Our entire science of classification would be moot. And these aren't even soft sciences, they are also hard science.
Though if you specify the basis for your classification, then I think the classification itself can become a testable prediction. You might have a category for "animals with features X and Y", because you've only ever found animals with both. This is implicitly the testable prediction that those two features will occur together. If you found a black-and-white-striped horse-shaped carnivore, then the biologists would surely consider something to have been falsified...
If you make a prediction that by its nature cannot be tested, you really can't do anything useful with it.
They might do one day.
But if we weren't willing to talk about them before they were testable then nobody would be inspired to develop the necessary tests.
The fact that they're an essential part of the scientific process makes them science as far as I'm concerned. If it helps make up your mind, plenty of people called scientists spend time thinking about them.
Some parts of quantum theories are inferred, but they’re inferred from data collected in reproducible tests. Also, any part of quantum theory that hasn’t been confirmed by experiment has scientists working to do just that. If something was untestable and thought to never be testable in the future, then it wouldn’t be considered science.
I genuinely think you’re confused about what science is. It is entirely based on empirical evidence. Broadly speaking it is one source of truth and knowledge, and while there are other sources of truth and knowledge that are also valid, being valid alone doesn’t make them science.
Psychology is probably a decent example here, as it has come up with a few tests that are reproducible (like IQ tests), and even manages to make some predictions given large enough sets of subjects, which is the only reason it's considered any form of science at all. Philosophy at the other end of the spectrum tends to almost never make any measurable predictions at all.
Empiricism is one way of providing evidence for a true statement, but -- from a philosophical perspective -- it is not obvious it is the best or only way. There are some things that are true that are simply beyond the ability of anyone to see or prove. That's supposed to be what philosophy is about (as opposed to psychology)