What's nice about building databases on compositions of CRDTs is that if you can validate the individual CRDTs via automated testing, you have a very high degree of confidence the composition of those CRDTs will do something similar.
No one's arguing that a Jepsen test shouldn't be done. Just that it'll probably be very different in character from more invented industry technology.
Yes; if you can prove a system was built atop an academically sound paradigm, it's more likely to adhere to the guarantees that paradigm is intended to provide than a system built atop an unproven paradigm. Kinda goes without saying, though. :P
Unsnarkily, I agree that implementation and composition of CRDTs is comparatively straightforward compared with other approaches people have taken. But if correct behavior is a requirement, full testing is too, regardless of how easy it -should- be in theory.
No one's arguing that a Jepsen test shouldn't be done. Just that it'll probably be very different in character from more invented industry technology.