Resolves #266 and improve saved log likelihoods and priors

• For the emcee, ptemcee and cpnest samplers. Adds the log_likelihood and log_prior per sample to the posterior (for CPNest this was already there, but was named differently).
• For all other samplers, the log_prior is calculated after the run based on the samples.
• Modified the prior.PriorDict.ln_prob method to include an axis argument

Closes #266 (closed)

bilby/core/sampler/cpnest.py

@@ -84,6 +84,8 @@ class Cpnest(NestedSampler):
             out.plot()
 
         self.result.posterior = DataFrame(out.posterior_samples)
+        self.result.posterior.rename(columns=dict(
+            logL='log_likelihood', logPrior='log_prior'), inplace=True)
         self.result.log_evidence = out.NS.state.logZ
         self.result.log_evidence_err = np.nan
         return self.result