- simulation is very sensitive to delta_p, like thermal flow in Theisen2021
  - do parameter study
- tau = 0.05 still not fully converged, needs some refinement, other cases look fine

- Git protocol apparently needs an user account (although public repo -> strange)

- Plotting doesn't make sense in that case, refer to errors.csv instead

- If eps_w != 0, pressure is determined WITH constant
- rescaling leads to wrong result, esp. visible for finer meshes (thus the test cases worked)