This is far too high a standard for anything simply called "empathy".

The behavioral correlates of empathy are widespread among social mammals, and there is approximately nothing save hubris that indicates that most of that works differently in us than in them.

If we layer our exceptional ability to create narratives about hypotheticals (note--complex sequential planning is widespread among mammals, too) on top of that system, and want to refer to this storytelling-about-our-empathy as its own thing, it needs its own term.

Much like self-consciousness is not the same as consciousness.

If the claim is, "we must be able to tell ourselves stories about being in the place of a slave in order to realize that slavery is wrong", I strenuously object. Basic empathy is more than enough. Indeed, slave-owners seem to have needed to use narrative skills to override basic empathy in order to convince themselves that slaves aren't really to be empathized with.

If the claim is instead, "humans did not have basic empathy until ~2k years ago when it arrived with a sophisticated accompanying narrative and we can consider the whole thing just 'empathy' (in humans)", I again strenuously object. We see behavioral correlates of empathy throughout our closest animal relatives, universally across cultures (including those that have changed little over that timeframe), anthropological indications of care and concern far older than that, and so on.

If the claim is that the best counter to our storytelling explaining why we can enslave people and it's okay is more storytelling imagining the roles were reversed, sure, that could be. If we're going to generalize messy, often-conflicting feelings into crisply-stated principles that help us make pro-social decisions that are win-win in the long run, this kind of universalizing imagination, and ability to share it with language, and build upon it, is quite important.