Factors of 2pi missing in Likelihood

[vhaasteren]

In signal_base.py the LogLikelihood class implements the likelihood. This likelihood is not including factors of 2pi that should be there. This does not affect many analyses. However, the fully marginalized likelihood / evidence will miss those factors of 2pi. That means the evidence is incorrect. This difference is quite large, since there is a log(2 pi) contribution for every data point.

This is not a normalization matter, the likelihood is normalized with respect to the data. All those factors of 2pi need to be there.