Kernel panics should reboot after a while
High availability systems shouldn't stuck on kernel panics indefinitely until another layer/person reboots the system.
At least, the system should still reboot after a while if init exits in an unusual manner.