fix the Jacobian at point of quad_3d_9

[vryy]

Description
One Jacobian function of quad 9 (3d) is wrong

Please mark the PR with appropriate tags:

• Api Breaker

Changelog
Fix the Jacobian function at a point. This is important for boundary conditions on hex mesh.

[RiccardoRossi]

makes sense to me. could u add a test?

[RiccardoRossi]

also i would check if the same problem also exist in the quad_3d_8 and quad_32_4

[loumalouomega]

strange, we had tests, but maybe because the tests for 3D geometries were the same as 2D geometries we did not notice. @RiccardoRossi I am quite sure Quad3D4N is OK, I use it a lot

[loumalouomega]

also i would check if the same problem also exist in the quad_3d_8 and quad_32_4

I have checked and these two are OK

[RiccardoRossi]

@loumalouomega apparently if we just flip the normal for Quad3d4 the determinant is not computed correctly. @philbucher could u comment briefly? i think you bug may be connected to this one

[philbucher]

@RiccardoRossi and I were briefly discussing this morning and it might be that the computation of the det of the Jacobian is also not correct in Q3D4N IIRC

there is an issue about the flipped normals => #7358

[loumalouomega]

@RiccardoRossi and I were briefly discussing this morning and it might be that the computation of the det of the Jacobian is also not correct in Q3D4N IIRC

there is an issue about the flipped normals => #7358

What is wrong?, there are tests