Add benchmark test for "heat transfer during the melting of ice around a horizontal, isothermal cylinder"

[agzimmerman]

Based on this paper: https://www.researchgate.net/publication/226804783_Heat_transfer_during_the_melting_of_ice_around_a_horizontal_isothermal_cylinder

The current enthalpy-porosity model should be enough for a nice qualitative result. A nice quantitative result, and perhaps easier set up in general, would first require phase-dependent heat capacity and thermal conductivity (for which we have issue #37).

This would be our first demonstration of an interesting geometry.

[agzimmerman]

With FEniCS one can easily construct the geometry by subtracting a circle from a square and mesh the geometry automatically with mshr. I don't think Firedrake has anything like mshr, since it focuses on it's utility as a FEM library and lets you import meshes from other software.

So the first task would be to create a procedure for generating meshes with different uniform refinements, perhaps by using fenics.mshr, or with any other tool.