Declare a system worker for generic training dataset

https://redmine.teklia.com/issues/10032

Generic Training Dataset is a worker used to create artifacts for datasets and mark them as Complete. We did not originally expect that a unique generic worker would be used to prepare datasets when we implemented datasets, but that's what's happening now. For example, the Crosslang tutorial uses that worker for both the segmentation and transcription training.

This tutorial expects that this worker has already been setup on the instance, but does not mention what to do when it isn't, so this only really works on our own instances. We can declare this worker as a system worker to let arkindex update_system_workers do this instead of end users, to make training models a little easier on other instances.