mpz_nextprime deficiencies

[Zimmermann Paul]

       Torbjörn,

>   this is indeed a function I'd also like to have (with a different name
>   due to upward compatibility, maybe mpz_nextprime_reps).
>   
> I have been considering an interface (to both probab_prime_p and
> nextprime) with a "strength" parameter instead, to allow us to change
> underlying algorithm.
> 
> Such a paremeter would be some sort of likelihood of error, but since we
> probably don't want floating-point parameters, an integer parameter x
> with p = 2^{-x} being the likelihood of error might be better.

good idea!

Paul