GitLab

simplify the model by always using circuit coupling (current source is explained in electromagnet.pro)

Impose homogeneous Dirichlet condition on the top and right surfaces, instead of homogeneous Neumann. It's a bit more natural, amd it matches the small example in the getdp documentation.

(this also makes the R-L system transient shorter, which makes the voltage-driven time-domain
results easier to understand)