MinimumEffectiveSamplesLikelihood is incompatible with update to gwpopulation

Our recent update to gwpopulation has log_likelihood_ratio eventually call _compute_per_event_ln_bayes_factors which now expects a tuple to be returned (since return_uncertainty is True by default. However, _compute_per_event_ln_bayes_factors is overwritten by the MinimumEffectiveSamplesLikelihood and only returns the per event ln BFs, meaning the log_likelihood_ratio does not get enough values to unpack.