Improve efficiency of edgeIndices function

Currently a `Set ( Int, Int )` is used to track unique edges, but this involves allocating (and comparing) a lot of tuples. Instead, build up a `Set Int` where each `Int` is a packed edge index equal to

```
lowerVertexIndex * numVertices + higherVertexIndex
```
Then, once that set is constructed, use `Set.foldr` to iterate through the set, unpack each `Int` back into an `( Int, Int )` using `mod` and `//`, and accumulate these values into a list. This should maintain the property that edges are returned in sorted order with lower vertex index first and higher vertex index second.

The issue identifies the edgeIndices function as the entry point; locate it in the repository and inspect its current Set (Int, Int) construction. Apply the packed-index approach described, then verify that returned edges remain sorted with the lower vertex index first and the higher vertex index second.

Written by the indexing model from the issue text.