PowerLaw prior does not check if prior minimum/maximum are valid

---------------------------------------------------------------------------
ZeroDivisionError                         Traceback (most recent call last)
<ipython-input-58-1f15b00eece0> in <module>()
      1 grid = np.linspace(0.1, 5, 100)
----> 2 plt.plot(grid, p.prob(grid))
      3 plt.xlabel('value')
      4 plt.ylabel('probability')
      5 plt.savefig('prior_1')

/usr/local/lib/python3.6/dist-packages/bilby/core/prior.py in prob(self, val)
   1016             return np.nan_to_num(val ** self.alpha * (1 + self.alpha) /
   1017                                  (self.maximum ** (1 + self.alpha) -
-> 1018                                   self.minimum ** (1 + self.alpha))) * self.is_in_prior_range(val)
   1019 
   1020     def ln_prob(self, val):

ZeroDivisionError: 0.0 cannot be raised to a negative power

Here my self.minimum was 0. We should catch this earlier.