It seems like you have a lot more confidence than I do that we know everything the WIV was working on? I agree that SARS-CoV-2 is unlikely to be derived (naturally, artificially, or otherwise) from any known virus, including RaTG13; but a distinctive characteristic of the WIV's research was their sampling of novel viruses from remote bat caves that no other humans routinely entered. That presents a pathway for emergence of novel, naturally-evolved pathogens beyond what normal agricultural etc. activity already risked, whether through a researcher infected in the field or through a literal lab accident back in Wuhan.
So specifically considering their group, the novelty of the pathogen doesn't seem to me like significant evidence against lab origin. The public focus on gain-of-function research actually seems excessive to me--while I do think it's possible that the origin of SARS-CoV-2 involves lab manipulation (e.g., genetic engineering or serial passage) of the virus, the WIV's library of novel natural viruses seems to me like at least as big a part of the danger. In any case, their database went offline on Sep 12 2019, around when that genetic clock suggests SARS-Cov-2 began spreading in humans:
https://www.researchgate.net/publication/349073738_An_invest...
That timing could be coincidence, and their flimsy explanation for its continued unavailability ("hacking attempts") could just be the reflexive secrecy of an authoritarian regime. It's a second weird coincidence, though. The WIV has claimed to have published everything they have to GenBank; but Deigin et al. were able to assemble a novel Merbecovirus from contamination of agricultural datasets from a team that shared sequencing equipment with the WIV:
https://arxiv.org/abs/2104.01533
That Merbecovirus obviously has nothing to do with SARS-CoV-2; but if they had one unpublished virus, it seems reasonable to wonder if they have more.
If you're going to extend the field of zoonoses to include events that didn't become major pandemics, then it seems like you should extend the field of lab leaks too? But let's say negligible probability of lab origin is p 100 (which per above I don't see).