It will be great if we can define the global dataset and use them across multiple bionic rules.
Thanks for posting this. One problem in this approach would be, if you had a source as another dataset then the field mappings will go for a toss and it will take as much effort as to map each field to the new input dataset. This will only work when you have a fetch task that fetches from MDA or SFDC. Does it sound like a huge limitation or you are good with fetch tasks alone?