feat: Adding Concrete NetworkDF

The network structure can provide powerful insights when modeling social interactions.

The agent’s attribute table already encodes nodes and node features. Meanwhile, the Network DataFrame would encode the relationships between nodes along with the attributes of these relationships. This setup should also allow for compatibility with other spatial structures.

For PyTorch integration, Ibis offers a [to_torch](https://ibis-project.org/posts/torch/#execute-the-query-and-convert-to-torch-tensors) command that simplifies tensor conversion. Additionally, we could introduce a command to retrieve a tensor-based representation directly, making it possible to leverage libraries like [PyTorch Geometric](https://github.com/pyg-team/pytorch_geometric) or [DGL](https://github.com/dmlc/dgl) for advanced relational analysis.

The issue does not name files, tests, or entry points. Start by locating the existing agent attribute table and spatial structures, then review how Ibis and the proposed PyTorch conversion relate to a Network DataFrame; done should include a defined relationship representation and its intended tensor integration.

Written by the indexing model from the issue text.