It is exactly what I am suggesting.
Many of the more complex models available here include their own designed support which was built-in to the overall design rather than uses the automatic support generated by the slicer. The designer usually knows more than the slicer about the model.
What I am suggesting is the equivalent for the void issue you speak of.
As the designer, you are capable of creating far more complex shapes than the slicer is able to achieve with primitives to meet your specific needs.
You would design a part specifically for it to meet the needs of the voids you originally discussed and the tell the slicer what infill patterns and/or percentages to use. You are creating the ability to fine tune those settings by have a complex shape rather than a primitive.
It is an advanced solution to what seems like an advanced problem of fine-tuning.