Add unit tests for d=0 in lazy sets

[mforets]

We should review that for d=0 the support vector returns a point in the set, or to consider the special cases that raise an error otherwise.

[schillic]

I reviewed this in #73.

d=0 should work as expected for the following base types:
BallInf, Ballp, HPolygon, HPolygonOpt, VPolygon, Hyperrectangle, Singleton, Zonotope
(and of course for all "operation" types)

For the following types we have to revise:

• Ball1 uses the center but changes one coordinate to 0. We should guard this change by checking if the direction entry is 0.
• Ball2 returns the origin. Instead we should return the center.

@mforets: Do you agree with these changes?

For the unit tests it will be helpful to have #68.

[mforets]

In some set operations we specifically say that the results depends on the wrapped sets, e.g.

LazySets.jl/src/LazyOperations/CartesianProduct.jl

Line 147 in 2423c0d

function ρ(d::AbstractVector, cp::CartesianProduct)

In other cases the check is explicit and an assertion error is returned, e.g. for Bloating,

LazySets.jl/src/LazyOperations/Bloating.jl

Line 92 in 2423c0d

function ρ(d::AbstractVector, B::Bloating)

---

In other cases as mentioned in this comment the operation returns a hard-coded value eg. the origin.

[mforets]

I think consistency across different types for this corner case would be nice for the next major release, so I'm adding this to the v2.0 milestone.