mpn_sqrtrem{1,2} - patch for pure C implem

[Adrien Prost-Boucle]

> The only branch is for the final correction at end of mpn_sqrtrem1.
> I tried with the previous mpn_sqrtrem1 version, which has a condition,
> and with an unconditional code that needs 2 multiplications.
> 
> My version with unconditional correction had same speed.
> My version with current mpn_sqrtrem1 correction is slightly faster.
> So it's the "core" that is faster.

So the floating-point sqrtrem1 the I proposed before should be a bit faster
is using the correction with condition, instead of the correction with only
unconditional instructions.

Adrien