[Rivet] Unsuitable error checking

[David Grellscheid]

> I also don't understand how it would waste one *day* of anybody's time
> to see that a job crashes and then submit it with stable Higgs or the
> analysis disabled.

My high-stats trial runs take approximately the time I'm out of the 
office overnight. Send job in the evening, go home, expect to look at 
plots in the morning. If they haven't run properly, I can of course 
re-send it right away that morning, but it will only be done by the end 
of the day, when I've left => 24h lost  :-)