Add high-level abstract coupling restriction

[pjaap]

This may be much more user friendly

[chmerdon]

Nice, how about using the normal vectors on the boundary regions to define the opposite function (assuming that the boundary regions are hyperplanes not necessarily orthogonal to an axis)?

[pjaap]

Uh, this is a brilliant idea. For this high-level restriction I assume that the boundary regions are hyperplanes and that the axis is orthogonal to the boundary regions.

For the general case, the concrete constructor with the coupling matrix is still there and should stay.

I'll push an update today :)

[pjaap]

Alright. Now there is no need for an axis any more.

The give_opposite! function is pretty tricky from a mathematical POV, but it can be implemented efficient. This does not seem to slow down the computation of the matrix.

[pjaap]

I am happy now. New example, new features :)

[chmerdon]

Nice! I think it can be merged. (I am in favor of renaming PeriodicBoundary to PeriodicElasticity in Example212 and Example312, though)

[pjaap]

like this?

[pjaap]

damn, classic mistake with the filename. We should definitely have an "exclude" list an parse all Example*.jl files in the examples folder.

But now merge?

[chmerdon]

Ok, ready to merge I think