Change Sigma*0 decay model to PHSP to fix EvtGen errors

In MC production, found that this was failing at a rate of 4% of jobs as reported here. I confirmed this in local tests over 10k events, where I ran 10 jobs of 1k events and 3 failed. This implies 1 failure every 2800 events. As suggested by Michal, I changed the Sigma*0 decay model from PYTHIA 0 to PHSP. Tested this change over 20k events with zero failures observed.