Solving `DFTProfile` for binary mixture with `x1 = 1` results in singular matrix about feos

[prehner]

The density solver cannot handle components with x=0 because their total chemical potential diverges. Even though a physical solution exists, it cannot be found because the code does not cancel out the diverging ideal gas contributions.

Some thoughts:

• maybe a method to convert PhaseEquilibriums between different component sets could be useful (i.e. add components with n=0 or remove all components where n=0)
• Solving density profiles for one (or more) non-zero components and then adding a zero component can be an interesting case because it gives rise to an inhomogeneous Henry constant and should not suffer from the same problems that the solver has with empty components.

[prehner]

This problem is solved in #60. However, any solver with log=True will still not converge as it will inevitably call log(0). So a solver has to be passed explicitly. In this case both

DFTSolver().anderson_mixing()

and

solver = DFTSolver().picard_iteration(max_iter=3).newton()

converge.