H2O Predictor Nodes to report prediction variance for conformal prediction

To facilitate the use of conformal prediction in combination with H2O-based machine learning it would be nice if the H2O predictor nodes could report the prediction variance, similar to f.e. the Random Forest Predictor nodes. Conformal prediction needs these prediction variances for the calibration step.

If this would be available it would be straightforward to use the H2O AutoML Learner nodes in combination with conformal prediction.