still bug in LF with nonlinear flow

[norihiro-w]

Bugfix from John after #51. TODO note for myself. I will soon update the code

I think that there is something missing from your patch: you evaluate k_rel correctly, but this term should then be multiplied with mat[i], like this:

            for (size_t i = 0; i < dim * dim; i++)
            {
                mat[i] = tensor[i] / mat_fac * perm_effstress; // AS:perm. dependent eff stress.
                mat[i] *= k_rel;    //JNEP
            }
            break;

With this change the nonlinear term is activated properly, according to my first tests.