adding ElementCustomIonPotential + new GP example using it (with doc)…

[gkemlin]

• new element for to generate AtomicLocal term with custom potential
• new example of 1D GP with SCF and forces using ElementCustomIonPotential with docs
• small typo in 2D GP example
• modified Model.jl to detect 1D/2D models and automatically disable symmetry

[antoine-levitt]

Also should that custom type be in the example file rather than in elements.jl? I think it makes sense since it's so simple. Also if in the DFTK source it poses the question of should Coulomb be a particular case of this, which I'm not sure we want to bother with.

[gkemlin]

I think that it's only 18 lines of code at the end of the element.jl file so having it implemented in the code or in an example doesn't change that much. In the former case, users only have to call ElementCustomIonPotential(Z, pot_real, pot_fourier) to implement their potential without thinking about this Element structure and it might be a good thing. But if you prefer the other way round, it's good for me.
Keeping Coulomb as a particular Element might be better ? This custom type is more for "exploration" whereas Coulomb has real interest in practice.

[antoine-levitt]

I think it's better outside: better to have a small code base that makes simple extensions easy than have all the extensions in the code.

[mfherbst]

Nice example. Yes I agree no need to put this into the main code. Maybe at some point when we find we need it more often.

[mfherbst]

No comma after that.