WIP: Refactor for more scalable design

[Roger-luo]

Based on #13

[Roger-luo]

Hi, guys I'm thinking about how to write a general edge iterator based on unit cells, I'm wondering implementing it in this way is sufficient:

1. unit cells provides an edge iterator that iterates its effective edge given its position (e.g for square lattice, its unit cell location can be (1, 1)), hypercude can be (1, 1) as well.
2. we iterate through given geometry (such as Plane, Sphere etc.) at each iteration state, we iterate the unit cell so we get a set of effective edges generated from unit cell

However, this will make our previous interface change a bit, e.g instead of saying we are going to have a 4x4 square lattice, we will say we are going to have a 2D plane of size 4x4 with square unit cells (which means there are 4x4 square unit cells)

Since I think it is harder to generate proper size for unit cells than just provide each unit cell's position and check if the geometry fits given unit cell.