The plate does indeed stick very strongly against the bed. The magnets are fine.
The most plausible explanation I have is that someone stuck their hand in the enclosure for some unexplainable reason and the bed hit their arm or something. The only problem with this theory is, there is zero reason for anybody to do that. But I did start the print early enough for a few people to still be at work when it happened.
Other than that, I've taken to using glue on the plate lately, because I've had adhesion problems. Maybe some leftover glue turned liquid-ish and degraded into some sort of lubricant with the heat (I'm printing PETG with the bed heated at 85C) and it seeped between the bed and the plate. I've never felt it was ever slippery though.
Or it was some leftover water that steamed over between the bed and the plate and lifted the plate for a second, air-hockey stylee.
None of this seems very likely though. So I cleaned the bed and the plate real thoroughly, installed the plate tight against the registration pins, upgraded the firmware to the latest for good measure, slathered glue super-carefully all over the plate, making sure it didn't go overboard, and started the same print once more. I'm watching it remotely through the webcam I installed in the enclosure and it seems to be going fine.

I do have a camera but it doesn't record.
If it was a nozzle hit, it was a violent one: this was 8 separate parts printing on the same plate and all 8 parts were similarly shifted - meaning the plate itself had shifted underneath, or both motors skipped or lost their origins. So I don't think the nozzle hit anything.
The filament is PETG, but I don't think it matters. The prints don't lift off the bed because I use glue.