where exactely do you see problems in your data model?
The MeasureTable should be a data island table, not linked to any other table. If the Measure field name conflicts with other field name, creating an unwanted key, rename the Measure field name in the MeasureTable to any other name you like.
Basically, you can only filter on table dimensions based on a field in your data model, this is why you can't filter on valuelist() synthetic dimensional values.
The problem is exactly the definition of island, i.e. the fact that is not connect to other entities in the data model.
In the example, I could not choose a worker to have his/her partial results in the measures used as dimension with an island.
I'd really like ti have the result of a valuelist(), measures as dimension, whom could be influenced by other dimension. In addiction, I'd like to use as "filters" the measure-dimension created by the hypotetically valuelist(), for further analysis.
The question is rather tricky, because the real meaning is that I do not want it working using whatever medium, I'd like to know if it works, and I could accept a workaround like your, but it should preserve the features of the valuelist() way.
I've tried your solution, and it does not work, unless you add an aggr() (If I have read it well, if not, my bad).
Measure = 'Sum'
However the point is that if you choose a worker, the result of the previous formula is not affected by the selection, becoming a dimension. The valuelist() makes the measures as dimensions, and the results of the measure-dimension affected by the filters as in the example, refreshed with your solution (added at the question).