should work. If I add set keep invalid mesh on and examine the resulting mesh, it seems to show that the curves in the fault planes are correctly connected to the surrounding tetrahedra. The error may be a false positive.
If you have time, you could try setting the “keep invalid mesh on” and check it in your analysis code.
Otherwise, the work around is to separate out the fault planes into an independent sheet body and then add the triangles with tetmesh respect.
reset
open "~/Downloads/cubit01.cub"
separate surface 66 64 61 62 63 65 67
surf in body 20 size 1000
surf in body 20 scheme trimesh
mesh surf in body 20
vol 19 tetmesh respect tri in surf in body 20
vol 19 scheme tetmesh
mesh vol 19