I think the physics of the situation don't make a barrier feasible: a derailed train going >100 mph is going to transfer a lot of energy to any kind of barrier it impacts, which in turn might exacerbate the situation (by spreading debris).

I think these kinds of accidents are largely mitigated by rail defect monitoring. I know rails in the US are equipped with defect detectors for passing trains; I'm surprised that a similar system doesn't exist for the rails themselves. Or more likely, one does exist and the outcome of this tragedy will be a lesson about operational failures.

I’d rather they spent the money ensuring no trains ever left their tracks rather than halving the destruction if they do.

There was a switchover which made the derailed cars of the first train move into the track of the second one, you can't have a wall there anyway.

More practical but still probably unnecessary is having the planned “passes” be where the tracks are separated by some distance.

But that requires the trains mostly always being on schedule.

The 20-ton bogie was flung 300m. What do you expect the weight of a whole car to do to such a wall?