[Feature Request] TimeSeriesCloudPredictor known_covariates

[austinmw]

It seems that currently TimeSeriesCloudPredictor training already supports known covariates by adding these to the training data frame and additionally including predictor_init_args["known_covariates_names"].

However predict_real_time and predict methods won't accept the additional/necessary known_covariates argument that needs to be passed for inference as described here.

Please add as this is an important forecasting feature 😁