[Graph] odd outcomes using node_species argument of operators that generates graph

[chapuisk]

Describe the bug Operators to generate graph (such as small world, scale free and the like) make it possible to specify the species of the node, i.e. argument node_species. Doing so leads to diverse odd outcomes: more nodes than agent of species created, node (vertice) that are actually not from defined species, spatial representation do not match, etc.

To Reproduce Steps to reproduce the behavior:

1. Copy past this simple model example:

   
   global {
   
   graph<a,unknown> g;
   init {
   
       g <- generate_random_graph(10,10,false,a);
   
       write g.vertices;
       write g.edges;
   
       g <- as_spatial_graph(g);
   
       write g.vertices;
       write g.edges;
   
       ask a {
           write sample(g contains_vertex self);
           write sample(g neighbors_of self);
       }
   
   }

}

species a { aspect default {draw circle(1) color:#grey;} }

experiment xp { output { display main { species a; graphics edges {loop e over:g.edges {draw geometry(e) color:#black;}} } } }

2. Execute model
3. See in console that 
5. See error

**Expected behavior**
Being able to build a graph using available methods without having to go through another operator (as_spatial_graph), having to maintain the link between inner graph representation of nodes and edges and my node species, being able to display it right away, etc.

**Desktop (please complete the following information):**
 - OS: macOS Sonoma
 - GAMA version: 1.9.3

**Side note**
It is counter intuitive to be forced to turn generated graph into a spatial graph to be able to manipulate and display it

[lesquoyb]

I also noticed some weird code in graphs while cleaning up the code base (see for example here a property was used in some graph algorithms but actually never set) I think that part has been modified many many times and would benefit from some refactoring

[ptaillandier]

@chapuisk to be tested, but it should be fixed