[1] You're not _really_ bounded since you have blocking calls to the underlying allocator (usually somewhere close to the OS, and even that can do more work than you might expect when you ask for a contiguous array), but it's still much cheaper than a bulk copy.
[0] It needs a refactor, but here's some version of the idea. https://github.com/hmusgrave/zcirc
[1] You're not _really_ bounded since you have blocking calls to the underlying allocator (usually somewhere close to the OS, and even that can do more work than you might expect when you ask for a contiguous array), but it's still much cheaper than a bulk copy.