Why are the filleted regions of the model a lot less dense with the polygons than the unfilleted surfaces?
And why does the maximum edge length get ignored in the unfilleted surfaces?
EDIT: Also, hello…
Bonus… it’s the same surfaces mirrored, yet Rhino fails to produce tangency within tolerance on one side but not the other…
because the filleted surfaces need more polygons based on your max. distance edge to surface, and needs to adjust the surroundings to it.
Simple planes will optimize areas that need less polygons.
What you can try is to use an initial grid quad >0, e.g. 8 or 16
Also see this page that illustrates how Rhino’s mesher works
Ehh… look closely at the orange arrows, because I’m seeing the exact opposite! 
Ah, is that the “refine mesh” option? No… so, how can I turn it off? I want control of my own optimization.
Simple planes is a checkbox, the lower left one in your screenshot.
Maybe a min edge setting and a bit higher setting for density can help too. If you can share the model I can make better suggestions. There is no one setting fits all solution, at least if you are looking for a mesh with reasonable resolution.
Did you try out different grid quad settings?
Oh, I’m so blind sometimes, thank you!
Will try to experiment with grid and see what happens.
The manual fix for now is to use “Remove knot” to get rid of all the duplicate control points located in the same place at this bad edge and leave just the last one.
