Global checkpointing dir. Default: Temporary

[PascalIversen]

new parameter for the run_suite, drug_response_experiment and DRPModel.train: model_checkpoint_dir. By default this is a temporary dir to prevent accumulation of too many checkpoints.

Closes #91

[JudithBernett]

Please try if this works with the pipeline:

1. On the server, pull this Git, create an environment, checkout this branch, install this current version with pip install -e .
2. Pull the current version of the nextflow pipeline, checkout the dev branch
3. Run the nf-core pipeline with one of the models that writes checkpoint files: nextflow run . --models [model-of-your-choice] --dataset_name Toy_Data --path_data [path-to-data] -c [your-config]
4. After the process TRAIN_AND_PREDICT_CV was run, find out to which work directory the results were written
5. Look in the work directory in verify that it doesn't look like this (i.e., that the checkpoints were deleted):