In a rollover accident, you also risk kids getting tossed around like clothes in a dryer and ejected from the vehicle if the windows are open or break. I can't say I think that is a better option. It is a risk vs reward scenario here. In a fire you can get out easier, in a rollover, you turn the bus into a drink mixer if it is going fast enough into the roll.
I’m willing to trust the people who designed them and the fact that they’ve continued with the same design over the thinking of a random person on the internet
The back of the seat if front of the students is higher and softer than the steering wheel in front of the driver.
As unfortunate as it sounds, in most accidents kids can't bounce around much and mostly hit something soft enough to keep injuries minor or at least nonfatal.
For a long time the numbers worked out that that was enough for most bus accidents to protect students, and that seatbelt costs would be better spent increasing safety at pickup and dropoff locations and increasing bus ridership numbers, since even without seatbelts a school bus is radically safer than being driven to school or walking in most places.
More recently, the numbers have started to say we should invest in seatbelts and making pedestrian routes to schools safer, since those would now make a more significant impact.