The FLASH in the secondary blast a few seconds after impact is a clear proof of a criticality event (first link : soldier report, second link : National Geographic documentary on the Gulf War, third link : video of a flash recorded after the impact).
This works also with missiles on the instant of the blast (undelayed), see for instance here for the Brimstone. No fuel explosion can create a flash like that, it’s too slow to compress air particles. Have you ever seen a car explosion generating a flash of light ? The use of uranium + beryllium / heavy water + sodium 24 / thallium 208 / polonium 210 – or a uranium – palladium – tritium mix in DU shells generates fast neutrons (from (a, n) or (gamma,n) reactions or a low energy nuclear reaction in the case of the tritium mix, see again links just above), plus an acceleration of neutrons at impact, melting down U238 from fission in the tip. In the case of missiles it is the direct fission of 238U from fast neutrons. In the case of tanks the delay comes from the continuation of (a,n) from the neutron source like beryllium and (n,2n) from heavy metals in the magma of uranium in the bunker / tank, which ends on a criticity excursion. The magma is full of neutrons from both (a,n) and (n,2n) and the cavity it has dug into the bottom of the tank allows for compaction of the magma, which thus becomes supercritical for a time. The amounts of energy released cannot of course be compared to a kiloton-range nuclear bomb, yet it is nuclear.
The fission products, as always, should be those of the 238U and also of the heavy alpha emitter used, since any alpha emitter is likely to fission (actually the emission of an alpha particle is sometimes defined itself as a spontaneous fission). Fission products WERE actually found in Iraq (the nuclear explosions of the Cold War were used to scapegoat).