turek-hron-fsi3: Add FEniCS as solid participant

I'm adding this issue as a summary of the current state and collection of useful resources, because it just took me a bit of time to reconstruct what has happened to this case during the last few years.

1. @richahert computed the FSI3 case with FEniCS in his thesis, section 5.2. Important: He used a linear solver. The case was contributed via https://github.com/precice/tutorials/pull/38.
2. In https://github.com/precice/tutorials/pull/125 we decided to remove FEniCS as a solid participant from the case. Main reason: because the linear solver does not yield sufficiently accurate results.

Adding FEniCS as a solid participant would be helpful for cross-comparison to other solvers (this already helped us to debug the perpendicular flap case, see #250). FEniCS also gives very direct control over the time-stepping scheme being used, which might give us additional understanding for the case (e.g. in https://precice.discourse.group/t/turek-hron-fsi-with-openfoam-and-calculix/1001/14).

I marked this issue as a student issue, because it might be a nice thesis topic.

Start with the FEniCS implementation described in the thesis, section 5.2, then read tutorials PRs 38 and 125 to understand the earlier contribution and why it was removed. Use issue #250 and the linked discussion for comparison context; done means FEniCS is restored as a solid participant with sufficiently accurate results for the FSI3 case.