Update of interface towards AtomsCalculator 0.2

[mfherbst]

Fixes #5, fixes #9, fixes #10

Things I still want to do here:

• Run examples as part of test suite or add them to the docs (where there are run during docs generation)
• More testing of the examples
• Check the results struct returned by minimise_energy! is reasonable.
• Test if other Optimisation.jl solvers beyond Optim can actually be easily employed and have an example of that.

[mfherbst]

@cortner Let me know what you think about this once you get a minute.

It might not solve all the things you had in mind with your three issues, but I hope it gets us considerably closer.

[cortner]

I did a relatively quick skim. The overall structure looks great. Since this is early days and few (any?) users I think it's ok to merge without in-depth review.

I will try to help with next steps such as integrating the DOF manager, masks, the lie-cell dofs etc.