I meant to answer this but life happened and I somehow forgot this thread. Even though it doesn't really affect Bitcoin I'll give this a shot anyway.
I assume this is the same “Schnorr” as in “Schnorr signatures”?
Yes.
I think he only found a fast method for special cases. Just from the abstract you can see that he had to use a certain number
n to make the factorization work, and even then it only works for particular numbers. I mean
come on, he literally used 2
400 and 2
800 as examples. Anybody can factor those, actually it's as factored as it can get (400 or 800 2s respectively - 2 is a prime number).
I could not fully digest the algorithms but as the Medium article said, for this discovery to be significant, it has to work on RSA numbers, which have exactly two factors. You'd have to find an
n that factors it correctly so it's not very useful, given that there is no formula that can guess the right
n if it even exists.