LOO-CV when log-likelihood is constant for some data points

[avehtari]

@flo

can I simply omit the log-likelihood for this first data point and only include the loglikelihood of the remaining data points when using loo?

Yes, this is the logical thing to do.