Password hygiene had absolutely nothing to do with this breach. It was an upgrade.
Whether they should have caught it in testing: arguably, they could have caught it during development, depending on what caused the issue and whether it was dependent on environment. (E.g. might behave differently between a dev's laptop and between AWS or whoever their new cloud provider is).
But there are always endless things to test and some escapes will happen, though I would expect most focus to be given to high priority areas (such as this one!). Whether testing would look at the effect of two users logging in at exactly the same time? If you have professional testers, yes, fuck yeah I would expect them to consider that a site with millions of users might have concurrent login attempts!
But many organisations will hire bargain basement testers who will plod through a few obvious scenarios and stop there. MN - if you're paying under £50k for testers in London, you are not paying enough.
The other side of it: the fact they didn't roll back swiftly, suggests either a failure of comms, processes, or that they really hadn't planned to have to do a swift switch back and ran around screaming for 12 hours.
All the above is guesswork of course.