Fix crash for FEniCS-OpenFOAM perpendicular flap example with subcycling

[NiklasVin]

Addresses the crashes when using subcycling as described in precice/fenics-adapter#172.

The current implementation of solid-fenics.py operates directly on the dof vectors of u, a and v at one point. This also changes the values stored in the checkpoint which ultimately leads to the crash described in the PR above.
Updating values of a function object via assign(), does not change the values stored in a checkpoint, since the old dof vector of the function to be updated is overwritten by calling assign() .

[BenjaminRodenberg]

I think this clearly improves the readability of solid.py.

It also hints at a problem with the FEniCS adapter: From my perspective it should not be possible to alter a checkpoint -- intentionally or by accident -- as soon as it is handed over to the FEniCS adapter. Meaning: As far as I understand this PR cures the symptom but we can only solve the original problem with precice/fenics-adapter#172.

[NiklasVin]

Yes, it is true that it only cures the symptom. With this PR, I also wanted to show, that the current issue is easily avoidable from the user's side. And since I am not working that long with FEniCS, I am not sure how common it is to directly change the values by directly accessing vector(). If it is more of an exception, I wouldn't be that content with the deepcopy solution, since it is, in the rest of the cases, a bit pointless, but I don't want to be a nitpicker. So yes, the deepcopy solves the original problem for all cases.

[BenjaminRodenberg]

Looks good to me and definitely improves readability. Let's merge it.

Let's discuss the deepcopy issue directly in precice/fenics-adapter#172 because it is now not really related to this tutorial anymore.