Optimize grid updates

Hey, thank you very much for all your videos, I'm having a blast watching them!

I made a Pull Request [here](https://github.com/CodingTrain/Wave-Function-Collapse/pull/40) to make grid updates more efficient. Instead of watching all cells one by one hoping that there might be some change, we can start from a modified cell (i.e. options are reduced for this cell), and update the neighbours. For each neighbour that has fewer options than before, we continue and check its neighbours etc.
We stop when no update is possible anymore, and the system is stable.

The idea comes from Constraint Programming, where constraint propagation is a thing that can be done this way. Here the constraints handle the compatibility between adjacent cells, and when a cell changes (its options are reduced), it triggers the consistency check of attached constraints.

The next step will be to collapse a new cell, and check its neighbours etc.

This allows to solve much larger instances, since it doesn't consider untouched cells. Here with 100x100 cells for instance:

Start by reviewing pull request #40, which contains the proposed grid-update optimization, and compare it with the current update flow. Done means updates begin from modified cells, propagate to affected neighbours until stable, and avoid reconsidering untouched cells so larger instances can be solved efficiently.

由索引模型根據 Issue 內容生成。