Handle/test duplication

[johann-petrak]

For now this is mainly about injecting the duplication ID into the script parameters using some reserved feature name.

[johann-petrak]

The basic way of running within GCP works, and each duplicate gets passed the number of duplicates and its own duplicate id in the kwargs. Logging when running in the console with gcp-direct.sh also works (apart from #21)