underflow in vol_prefactor for large dimensions

[kbarbary]

For large numbers of dimensions, vol_prefactor gives 0. due to underflow:

In [8]: nestle.vol_prefactor(450)
Out[8]: 5.73e-322

In [9]: nestle.vol_prefactor(500)
Out[9]: 0.0

One should probably not be using nestle for dimensionality this large, but it would be good to fix this anyway.

[joshspeagle]

I actually just ran into this problem for for computing both the prefactor and the determinant in ~250 dimensions. The fix is a bit of a pain, but just involves moving everything to log-based arithmetic.