`is_inscribed` depends on order of vertices

[gh-kliem]

Currently the inscription test for polyhedra depends on the order of vertices:

sage: Polyhedron(vertices=[[-2,-1], [-2,1], [0,-1], [0,1]]).is_inscribed()
True
sage: P = Polyhedron(vertices=[[-2,-1], [-2,1], [0,-1], [0,1]], backend='field')
sage: P.is_inscribed()
False
sage: V = P.Vrepresentation()
sage: H = P.Hrepresentation()
sage: parent = P.parent()
sage: dic = {True: 0, False: 0}
sage: for V1 in Permutations(V):
....:     P1 = parent._element_constructor_(
....:         [V1, [], []], [H, []], Vrep_minimal=True, Hrep_minimal=True)
....:     dic[P1.is_inscribed()] += 1
....:     
sage: dic
{True: 18, False: 6}

The algorithm constructs a sphere around dim + 1 vertices in general position. The circumcenter is computed up to sign. Then, one vertex is taken to determine, which sign to choose. However, up to dim vertices might lie on the intersection of both spheres.

We fix this by checking distance from the circumcenter for all vertices of that simplex.

[gh-kliem]

New commits:

​3f9cc75

check sign of circumcenter using all vertices of simplex

[jipilab]

Looks good to me. Thanks for fixing this error!

Again, the pyflakes is fixed in another ticket.