Despite the complete internal rewrite, and the move to async iterators instead of streams and promises instead of callbacks, it actually doesn't break any of the existing API; it's fully backwards compatible (and still passes all the original tests aside from some internals tests).
The new features probably won't all be supported with the old API, but internals improvements will.
I don't expect that this rewrite is going to be accepted as a PR upstream, given how significant the changes are, but in the worst case I just publish it as a fork :)