Half-agree: "Blockchain" systems contain new and useful technology, but the useful technology is not new, and the new technology is not-so-useful. If we keep the useful stuff, we're basically back at regular old distributed databases.

The key blockchain requirement is allowing unrestricted node membership. From that flows a dramatic explosion of security issues, performance issues, and N-level deep workarounds.

In the case of a bunch of banks trying to keep each other honest, it's drastically simpler/faster/cheaper to allocate a certain number of fixed nodes to be run by different participants and trusted outside institutions.

One doesn't need to trust every node, just the a majority is unlikely to be suborned, and you'll know in advance which majorities are possible. The bank in Australia probably doesn't want or need to shift some of that responsibility outside the group, onto literally anybody who shows up with some computing power.