Bug in the evidence from merged runs

The evidence from merged runs is being over estimated due to this line.

To verify this, here is some example code:

In [19]: from scipy.special import logsumexp                                    
In [20]: a = np.array([11, 12, 13])   # Create fake Z's                                          
In [21]: np.sqrt(np.mean(a**2))       # Actual RMS value calculated directly                                       
Out[21]: 12.027745701779143

In [22]: lna = np.log(a)              # Take the natural log                                                
In [23]: np.exp(0.5 * logsumexp(2 * lna, b=1. / len(a))) # This is how the RMS should be calculated                       
Out[23]: 12.027745701779143

In [24]: np.exp(logsumexp(2 * lna, b=1. / len(a))) # This is how we do it currently                          
Out[24]: 144.66666666666666