But it isn't a crypto attack at all. Gpg is doing what it is supposed to do, and the overall cryptography is not effected. It's bad clients that are the problem.

It strips the PGP MDC and then uses a malleability attack on CFB or CBC to inject content, but it's not a crypto attack? You have a definition of "crypto attack" that nobody in the field uses.

To be fair, a very large part of this attack is not about PGP at all. That part is about abusing email-client rendering of partly encrypted messages, and abusing S/MIME to get message malleability again exploited via HTML.

Again, this is like saying that BEAST isn't really about TLS, because you need a particular combination of client features to exploit it. The two attacks are almost exactly analogous in this respect. But PGP has a cheering section, and TLS doesn't.

I mean that the first attack (having A PGP encrypted middle of an email that the client just expands to plaintext) and the attack on S/MIME have nothing to do with PGP mistakes in PGP at all.

The gadget attack on PGP is completely an exploit against PGP, but this publication also treats other attacks. At the very least, if you weigh this by volume of text, they focus a lot on pure client mistakes (first attack) and S/MIME (half of second attack).