TaskRegistry refresh

[kels271828]

If someone is calling onemod through Python instead of the command line, and they call run_pipeline multiple times in the same script or notebook, they can get an error like:

NodeDependencyNotExistError: Upstream node, 1822966715896605700, for node, 743241294815203502, does not exist in the dag.Check that every task has been added to the workflow and is in the correct order.

This is because the TaskRegistry.registry dictionary is saved on the class, and tasks from the previous workflow are not removed before running the next workflow. Simple solution is to reset the registry before running a workflow.