[patch] libstdc++: Limit Riemann zeta testcases on simulator

[Paolo Carlini]

Hans-Peter Nilsson wrote:

>Jesper mentioned that the patch and the observations apply
>*only* when running on a simulator.  I think it's appropriate to
>prune the amount of numerical churn for such targets.  Similar
>patches have been applied to other time-consuming tests in the
>libstdc++ testsuite.
>  
>
Yes, I agree. Can you check that we are following a consistent approach? 
Then the patch is accepted.

Paolo.