The initial question was whether cryptocurrencies have the capacity for consumer protection measures, not whether those measures are ready for prime time. I'm just arguing that when they're ready, they'll be more efficient than what we have, so we should encourage their development in the meantime (though maybe not in Libra's particular case).
Also, there are limits on how old a transaction can be when a bank goes and rewrites history. In my experience the limit is about six months. Transactions older than than are considered settled.
If this is a feature that people want in a cryptocurrency, it shouldn't be hard to achieve with smart contracts. The problem right now is just that you need a solid settlement layer before you work on features supporting the politics of rejiggering unsettled transactions.
Also, provided there is community consensus (this differs based on whether your currency is proof of work or proof is stake) blockchains can be altered after-the-fact to undo a threat. It happens:
https://spectrum.ieee.org/tech-talk/telecom/internet/ethereu...
It's just that for most currencies, it's currently a political affair that occurs at a risk to the stability of the overall system. But there are (what appear to be) good technical solutions to that (decred, for example, has a neat approach to post-fork-attempt stability https://medium.com/decred/detailed-analysis-of-decred-fork-r...).
As far as deciding whether a transaction ought to be settled in the first place, people are experimenting with some really interesting approaches (https://particl.wiki/learn/market/mad-escrow for instance).
It's probably not time to forget your bank password and switch to crypto, but if we want to eventually have good solutions to our fraud problems then we should be working to shape crypto into the system we want, not dismissing it as inflexible.