Thank you Daniel, this looked very promising! However, when the “cloth” had landed, it suddenly started “boiling”, like so:
Perhaps some parameter can calm this down?
Edit:
This MeshLineCollider component is interesting and a good start, although it tends to fall through the target mesh and also doesn’t seem to come to full rest. Trying to calm things down it tends to make the cloth either too stiff (won’t go very deep down in valleys and grooves), or if increasing the Load it tends to bleed through the target mesh.
So, if it the cloth could be more “heavy” and also more elastic (reaching closer to the bottoms of grooves and valleys) while at the same time not being able to “bleed into” the target mesh, then it will do a good job patching “hopeless” bad mesh surface, which would be very difficult to patch in any other way.
I have tried manipulating Load, Strength of LengthFactor (of the cloth), also the penetration Strength of the target mesh (MeshLineCollider), I also tried to modify the Smooth strength, but the cloth still wont come to rest, and also bleeds though a little bit in places.
But this is getting close, and I think this is pretty cool! 
I also tried to play around with the code in the MeshLineCollider, but I didn’t achieve much. I guess you’re right in that a triangle collider would be better at preventing bleed through. Increased cloth elasticity perhaps can be fixesd with existing components? (I have not checked that as of yet but I have that next on my list). For performance reasons perhaps one could lump together some elasticity and load into the same collision solver?
Anyway, thank you very much for your work, it looks very promising and can open many new possibilities wrestling with “hopeless” meshes.
// Rolf