[Rivet] Rivet crash when applying projection

[Frank Siegert]

After very strange debugging experiences* I now think that it is related 
to our optimization setting -O3. When I switch this to -O2 I can't 
reproduce the segfault that I was able to reproduce earlier.
Maybe the strange thing that James saw in Athena had something to do with 
the optimization as well?

I know that we use -O2 in Sherpa. Would anybody be opposed to switching 
Rivet to -O2?

James: Any chance you could try that for the Athena problem?

Frank

* Putting in one 'cout<<"nothing"<<endl;' makes the segfault go away, and 
valgrind doesn't show any errors in that case.