Add error computation for partitioned-heat-conduction OpenFOAM solver

[davidscn]

To help testing and comparing.

Related to #445

In our tutorials, with OpenFOAM - OpenFOAM, I get

Dirichlet:

Error metrics at t = 1s:
l2 error (sqrt(sum(err^2)/n)):		6.2733e-05
Maximum absolute error (max(err)):	7.72464e-05
Global absolute error (sum(err^2)):	3.93543e-05

Neumann

Error metrics at t = 1s:
l2 error (sqrt(sum(err^2)/n)):		6.24377e-05
Maximum absolute error (max(err)):	7.4964e-05
Global absolute error (sum(err^2)):	3.89846e-05

[MakisH]

This looks like something we can easily merge today.

Apart from the value of beta (0.2 instead of 0.3, for consistency with the rest of the tutorials), this is identical to the OpenFOAM-preCICE paper code.

[MakisH]

I merged develop into this branch. The diff looks a bit different just because of the new formatting config added in #468.

[BenjaminRodenberg]

@MakisH to help with archeology here: In #379 I changed everything to 1.2 according to the FEniCS book.

[MakisH]

ok, then the OpenFOAM-preCICE publication is still in synch with the old tutorials, and the new tutorials are in sync with each other. All good, thanks!