I wish fusion powered tanks could go crit too. 
The tanks all sound like they are powered by 2-cycle lawnmower engines, so this could explain why they don't go crit.
I think it should only ever inflict heat, because canonically:
There have been a number of cases of fusion engines being "over revved" and exploding with devestating force, but this is more akin to a boiler explosion than a true nuclear explosion. More often a destroyed engine will be punctured by weapons fire. Because the plasma is held in a vacuum chamber (to isolate the superheated plasma from the cold walls of the reactor; contact with the walls would super-chill the plasma below fusion temperatures), a punctured reactor can suck in air where the air is superheated. Normal thermal expansion of the air causes the air to burst out in a brilliant lightshow often mistaken for a "nuclear explosion" that is only a risk for unarmored infantry close to the destroyed 'Mech.
Doesn't even sound like a catastrophic explosion, just a burst of hot air.
Why are we getting EMP effects from these things?
Even a real thermonuclear explosion would do most of it's damage to large metal objects by heat, but this is a good point regardless.
I also think the blast should inflict only heat damage, but that it should dump so much heat on nearby vehicles that they stand a good chance of taking considerable damage from it. Half the stuff in the current blast radius should be near certain to shut down, those already near overheat should probably pop, and if you have a direct line of sight to the blast, you should receive at least a small amount of heat.
Going by this, why does the explosion even have a delay? If your reactor is breached, seems like it would be an instantaneous explosion.
This would be F**KING AWESOME ! The delay is lame, it gives you a chance to run away XD
But seriously, I would love to see nukes with a shorter delay, more damage, giving heat to the targets around it. AND this should depend on the nuked ´mech. A nuked Atlas shouldn´t make the same damage and heat as an Osiris . . . [But of course there are other priorities now
]
I think the delay should be variable. There are plenty of reasons why a delay might exist, and plenty of situations where I could see it as being near instantaneous.