[models.model] Explicitly set the datamodule batch-size to 1 to avoid issues when running in an combo-experiment

[models.model] Implement to() method to move all computing to the designated accelerator device, including losses and augmentations

Make code and tests flexible to the use of a pre-processed Montgomery dataset to speed-up testing where possible (closes #74)

During validation logging, lightning seems to append "/dataloader_idx_n"
to the key we define if multiple dataloaders are used.

This currently only supports a single validation dataloader

[scripts.upload] Make uploaded data a bit more obvious; Report size of uploaded material; Report URL of uploaded model/experiment; Move gitlab functions to dedicated module; Document dedicated module; Fix QA