Model ensemble pull request

[orivera2280]

Most changes are in the base_trainer and property_trainer files. Generally, model ensemble methods are indicated by if statements checking for the type of model (if it is a list, it is a model ensemble) or by checking if the "model_ensemble" parameter from the config is > 1. I also made changes to the task.py file to support predictions on ensembles. These are indicated by the checks for the type of the data loader (if list, then it assumes a model ensemble is being used).