New hypergraph (and hypergraph vertex) properties

[espottesmith]

This PR implements the quad clustering coefficient for undirected hypergraphs (DOI: 10.1063/5.0188246; see also https://github.com/Gyeong-GyunHa/qch) and defines the diameter of a hypergraph as the maximum distance between any vertices.

[pszufe]

Thanks for this implementation!
Here are the comments:

• It looks that you use Matrix{Bool} as the major type throughout your code. Matrix is non-abstract type and such code seriously limit interoperability - AbstractMatrix should be used instead
• Matrix{Bool} is a dense matrix while the incidence matrices are sparse. This will be a huge performance bottlenecks for larger hypergraphs.
• Hypegraphs.jl distinguishes vertex-hyperedge incidence value weight = 0 from a situation when there is no incidence. In the Bool representation you assume that h = Hypergraph(Matrix{Union{Bool,Nothing}}(zeros(Bool, 2,2))) will yield a matrix of ones(Bool, 1, 1) while this might make sense - it should be documented in order to to surprise the user.

In order to provide good performance I could propose (for your consideration):

• implement SparseHypergraphView{Bool} <: SparseArrays.AbstractSparseMatrixCSC{Bool, Int64}. (it could be perhaps SparseHypergraphView{T} where T <: Real)
• have your methods accept argument AbstractMatrix{Bool} and iterate the arrays elements differently depending how issparse(arg) resolves.

If we have such implementation this should be both fast with large hypergraphs and flexible.

[espottesmith]

Sorry for the serious delay.

How's this? I removed the use of the Bool incidence matrix altogether. Now I'm just checking if the relevant values are nothing or not directly from the hypergraph. Would you still be worried about performance in this case?

[espottesmith]

Hey @pszufe, bumping this

[pszufe]

Thamks looks great! I removed two commented-out lines as they got. by accident from your code.

[espottesmith]

Thanks @pszufe