You can’t define measures in the dataflows at all, what happens in the dataflow is only equal to what is going on in the query editor. That’s why the data set is still needed.
And my company is small enough that we have two developers for BI - myself and my colleague. And the models that he and I are using are very different - he needs reports around accounting data, while my reports focus on sales and customer data.
There is a small overlap in what we need, but we are both able to handle those few measures in our distinct data set. Currently, we each have only one or two datasets, and all of our reports are connected to one of the datasets we have created.
Example - I have two datasets, one for internal use, the other for reports that are ‘customer facing’
these two data sets are connected to multiple reports. But both datasets are ‘fed’ from the same set of data flows.