CE Slab signature

I think the CESlab signature needs a bit of rework. It currently forces the user to use the ase.build.surface function, which is not the only possible option - in fact I believe it's a little restrictive. One could for instance instead use the premade fcc100 etc. functions, which would be a lot more intuitive.

I realise you can just circumvent this issue by skipping the CESlab function all-together, as it's just building the slab for you from a bulk cell (or symbol), but then it seems a little redundant.

I guess this also relates to how we suggest people use CLEASE: One option is also to stop trying to build the primitive cell for you, and just say "buildprimitive cell, and throw it in a settings class, and we just assume that whatever atoms you pass in is primitive".

Thoughts?