Google don't like canonical links, it's a burden. And if you pay attention to search business, Google recently has some problems with canonical indexing.
Q2A register and login links does not have rel=nofollow tags, and Q2A developers underestimate this because whenever you browse a Q2A question page, a tag page, a category page, a user profile page, there is a bunch of dofollow canonical links.
Thou we can disallow /register /login in robots.txt, it's still good to add rel=nofollow tags in the first place.
The coding style of Q2A core is hard to add such a small trick.
Recently I have a new Q2A installation. Not many questions, but a horde of excluded pages because of register/login links.
When you have >1000 excluded canonical links, you will understand the matter. Much more painful if the number is 9.6k