Now Lemmy can implement anything but nothing could ever prevent blocked/muted user to create another account in order to continue harassment.
But it’s not specific to Lemmy and same with anything open on internet.
I think the only way to prevent such issue would be a system which would require to prove identity in some way in order to create a single account. But this is completely against the openness of a federated network.
Indeed, I was focusing mainly in the fact that there is not a easy solution to completely avoid this problem. Now indeed being able to mute, to ban or to “report” an user to his instance admin could be useful tools even if they are not the full solution.