[Rivet] Sherpa+Rivet: m_io->fill_next_event fails

[Hendrik Hoeth]

Thus spake Eike von Seggern (jan.eike.von.seggern at physik.hu-berlin.de):

> Yepp, there's a cross-section line with a nan.
> 
> Any suggestion why the error is nan or how I can prevent this?

I'll send you a patch.

-- 
"Two equal numbers are equivalent, even if they come from two
 different people one of which you don't like"
(MLM about the theoretical top cross-section, in a meeting
about theoretical uncertainties on the top mass)