Improved cache management

To prevent cache exhaustion with long running jobs each job can be run with a --prune flag. This will automatically prune cache state at the end of each successful job completion. By default this is set to true.