Lots of things "don't need" to be written in python, but they're written in python anyway. I expect a lot of blockchain institutional use cases will take a similar route: sure, theoretically they could use some standardized federated solution, but in practice blockchain tx fees will become low enough through layer-2s to be negligible and that's where all the infrastructure to help users verify what's going on, and use those verifications in other applications, will be built. NFTs are already an incredibly easy to use and verify type of certificate; their biggest weakness is that (because of limitations in technology that are completely temporary and will disappear soon) transaction fees are too high.
Like... why _not_ dump Merkle roots on-chain?