Add DigraphCartesianProduct and DigraphDirectProduct methods

[reiniscirpons]

This PR implements two types of digraph product as discussed in #226 for a collection of digraphs. DigraphCartesianProduct computes the Cartesian product of a collection of digraphs and DigraphDirectProduct computes the direct product of a collection of digraphs. It also includes two new attributes DigraphCartesianProductProjections and DigraphDirectProductProjections for storing projections onto the coordinates of vertices.

Since the vertex set of DigraphCartesianProduct(G, H) is the Cartesian product of the vertex sets of G and H, but the Digraphs package does not permit a vertex set to contain pairs of integers, only integers themselves, then we encode a pair (i, j) as i + n * (j-1) where n = DigraphNrVertices(G).

If D = DigraphCartesianProduct(gr_1, gr_2, ...) then DigraphCartesianProductProjections(D) returns a list of transformations, where the i-th transformation corresponds to a projection onto a copy of gr_i contained within D. The projections are useful as they allow for us to reconstruct (under some circumstances) the graphs used for creating the product, and also provide for a way of regaining the information of the coordinates of each vertex (which were lost when we decided to encode (i, j) as i + n * (j-1)).

To make sure my code works with mutable as well as immutable digraphs, I used DigraphJoin as a basis, changing only the moving parts responsible for creating the digraph and leaving the more technical bits untouched. It seems like this has worked, but I would appreciate if someone told me whether everything is sound with regards to mutability.

Not exactly sure about the time complexity. For two graphs I think it should be O(E_1 + E_2) for Cartesian product and O(E_1 * E_2) for Direct product, where E_1, E_2 is the number of edges of each digraph.

[wilfwilson]

Hi @reiniscirpons, I've been making some fairly major changes to the Digraphs package as part of introducing mutable digraphs and working towards Digraph v1.0. Unfortunately that means that I've created conflicts with this PR, so this will need to be rebased.

(However, I'm not finished making the big changes, so you should probably wait until next week before resolving the conflicts.)

[wilfwilson]

Should be safe to rebase and resolve conflicts now.

[wilfwilson]

I've done a preliminary review - I've made some comments relating to the Cartesian product stuff, but probably the analogous comments apply to the direct product stuff, so if you could make changes in both places that would be great.

[wilfwilson]

Thanks for the changes @reiniscirpons, I have a few more :)

[wilfwilson]

Good work @reiniscirpons, thank you!

[wilfwilson]

Whoops, it seems that GitHub has some key commands that I accidentally pressed, which meant that I accidentally merged this before the tests had finished. Thankfully I'm sure they were going to pass anyway!