@Stein 's work at Comput. Methods Appl. Mech. Engrg. 417 (2023) 116483 corroborates my idea to apply a vector Nitsche method via \int (α/h) (σ \cdot n) (τ · n) ds.
With respect to the symmetry term, I like his argument at Implementing Nitsche's method for perfect-slip BCs with a nontrivial normal vector - #11 by Stein , “You don’t care too much about the symmetry term”