Ok, i did some redesigning to avoid the avalanche, since slight differences in the pulse HEXfet bank could cause the HEXfet that avalanches first to be under duress.
Eventually had to dump about 1800 Amps using a pair of IRLB3036 driven by a series connected pair of 1.5KE18CA tranzorbs. The snubbing Nfets arallel pair causes the kickback pulse to be routed to the Source of the original Pulse bank.
The snubbing Vgs is protected by a 16V Zener (to ground) fed by a 5ohm from the 50V + triggering tranzorbs. Another 5 ohms in parallel with the 16V Zener and 5 ohm gate resistor grounds the tranzorbs for rapid turnoff of the gates (10 ohms to gnd).
The 1800 Amp snubbing pulse lasts about 1.5 uSec now. LTspice says the kickback is now 54V clamped. No avalanche in the original pulse bank. The 12V gate protection zener is seeing perhaps 350mA during the 1.5 usec pulse.