#1352 - Disambiguation for lazy intersection of polyhedra

[mforets]

Closes #1352.

Marking this as WIP since i consider it still under discussion.

[schillic]

I see your point about unboundedness, but as you said here we already have this issue in the other methods because we cannot statically detect boundedness of general LazySets.
The only alternative I see is to manually wrap the constraints in an HPolytope, thereby expressing boundedness. The user has to deal with the case that the intersection is unbounded either way.
So from a usability point of view I would say we can add your disambiguation.

[schillic]

Maybe we should assert boundedness of X in overapproximate_cap_helper (compare to #1517) or change the return type of Q accordingly.

LazySets.jl/src/Approximations/overapproximate.jl

Lines 360 to 380 in b451d0a

	function overapproximate_cap_helper(X::LazySet{N}, # compact set
	P::AbstractPolyhedron{N}, # polyhedron
	dir::AbstractDirections{N};
	kwargs...
	) where {N<:Real}
	Hi = constraints_list(P)
	m = length(Hi)
	Q = HPolytope{N}()
	for di in dir
	ρ_X_Hi_min = ρ(di, X ∩ Hi[1], kwargs...)
	for i in 2:m
	ρ_X_Hi = ρ(di, X ∩ Hi[i], kwargs...)
	if ρ_X_Hi < ρ_X_Hi_min
	ρ_X_Hi_min = ρ_X_Hi
	end
	end
	addconstraint!(Q, HalfSpace(di, ρ_X_Hi_min))
	end
	return Q
	end

[schillic]

I fixed the remaining ambiguities.