Update anisotropic viscosity cookbook with new parameters

[Wang-yijun]

We've updated the anisotropic viscosity cookbook following the changes made for a recent paper revision. The main structure of the code remains the same, while the constitutive equation between anisotropic stress and strain rate, the equations to compute the anisotropic parameters (F, G, H, L, M, and N), and the coefficients and intercepts in the equations have been modified.

[gassmoeller]

I cannot contribute much to the code changes (they look fine to me), but can you leave a comment with some additional information (for future reference) in this PR:

• What is the benefit of the new values? Where the old values wrong, or are the new ones just more accurate (or are they based on different theoretical models / experiments?)
• Do we need to update the figure of the cookbook?
• The velocity in the test is changing, but none of the other values are. Does that mean the code change is not changing the computation by much? Or is our test not sensitive to changes in the algorithm? If it is the latter: Can we change the test to fix this (e.g. run it for a longer model time)?

When these questions are resolved this can be merged.

[gassmoeller]

Oh also: The failing test should be fixed if you rebase this PR to a current main branch.

[Wang-yijun]

Hi Rene,

• What is the benefit of the new values? Where the old values wrong, or are the new ones just more accurate (or are they based on different theoretical models / experiments?)

The update on the equations and coefficients uses a more similar method to VPSC, which avoids false estimation of stress using the original MDM method in special cases (mostly normal deformations). The resulting Hill coefficients and textures have a better fit with VPSC predictions now.

• Do we need to update the figure of the cookbook?

The figure in the cookbook mainly shows the anisotropic stress tensors visualized using tensor glyphs compared with (isotropic) stress tensors. The updates in this PR do not change the figure with a visible difference, so I didn't include a new figure here.

• The velocity in the test is changing, but none of the other values are. Does that mean the code change is not changing the computation by much? Or is our test not sensitive to changes in the algorithm? If it is the latter: Can we change the test to fix this (e.g. run it for a longer model time)?

I've updated the test to reflect the change. The main thing with the old test is that it only produces the anisotropic tensor at time step 0, which is isotropic.