Commenters on other boards have noted how the gash in the Crystal's bow looks like it was made by the anchor chain being dragged across it at tension. It kind of makes sense that the anchor grabbed the Fitzgerald at one point and was pulled out. The chain cut across the Crystal's bow as the Fitzgerald continued on, eventually dragging the Crystal's bow to the right. Probably dragged the Fitzgerald around a bit before ripping away. No idea if that's even possible - I'll leave that to mariners for comment.
The only reason I bring up the 'anchor snagging' situation above is in the one (unlikely) case where nobody was on the Crystal's bridge, the collision occurred and pushed/dragged the Crystal's bow 90° before the ships untangled. The auto-heading would have presumably steered the Crystal back on it's previous heading (with an anchor in tow). That would explain the AIS course changes. The crew would have made it to the Crystal's bridge afterwards and tried to figure out what happened, maybe not seeing the damaged bow immediately.
Mariners say the above is unlikely because a commercial ship's bridge is never unmanned in a busy shipping lane and there are various collision warning alarms that would have sounded if someone was 'asleep'. There's also emergency stops that kick in if a collision is detected. So there's no way, according to merchant mariners, the Crystal would just run on 'autopilot', ram another ship and resume course like nothing had happened. That sounds reasonable to me.
That still leaves the peculiar track of the Crystal unexplained. If it was under manual control and the ship did turn at the last minute but was unable to avoid the Fitzgerald, then they would have stopped immediately after the collision. I can't imagine any scenario where they would have resumed course/speed for nearly a half an hour before deciding to turn back - no matter what was going on on the Crystal itself.
So as unlikely as the first scenario is ('autopilot' ramming Fitz then resuming course/speed), that one - at face value - seems to explain the AIS track. Once again though, I can't possibly claim to know anything about what these crews would do in this situation - there was probably much more going on that we don't know yet. Neither explanation makes much sense, so I'm inclined to think both are wrong (or incomplete as we understand them so far).
"...And if they had to know they collided, why no distress call for an hour?..."
This may have been one of those 'lost in translation' things. If you read the articles carefully, the Japanese Coast Guard says they were not notified until 2:20. It doesn't even say if the Crystal's crew notified them. One of the earlier reports said the Crystal notified their shipping company first for direction, and then the shipping company itself eventually notified the Japanese CG after verifying with the USN. Too many inconsistent reports for now - I think we'll have to wait to find out how the 'reporting' chain actually happened.
The media makes it sound like the greatest crime of all here was that the Crystal didn't report the collision the minute it happened. FWIW, an aircraft pilot's priorities in an emergency are to aviate (keep the plane flying), navigate and THEN communicate - in that order. You don't grab the mic and start blabbing away to someone the second your engine falls off. That's not the immediate priority. I can't imagine a mariner would act any differently. In the event of a collision, they are worried about other things far more important than 'notifying the Coast Guard' who are too far away to be of much help anyway. I wouldn't read too much into the 'one hour delay' until we have a better picture of what happened.