Gyroid shouldn’t really be significantly more wasteful than rectilinear. A little slower and messier due to the accel’s, but only marginally more material.
Keeping my thumbs pressed and 
The real test for the infill will come at the top where it needs to support the top layers. You’ll need sufficient infill density and top layer count to hide bridge sagging. At the same time, shallow angled tops can suffer from edge curling so top layers can benefit from a tighter infill.
Well, I understand the sentiment, but grid for example is not really an option with todays speeds. So there are physical limitations we need to be aware of when slicing.
Incidentally, you did originally set both the “Internal solid infill pattern” and “Sparse infill pattern” to Rectilinear, right?
The reason I am asking is that your problem occurs with the sparse infill rather than the solid infill. The sparse infill uses grid as a default…
& 