[Feature]: Implement indexable properties for Collection attributes using the descriptor protocol

Currently, many properties in our codebase return complete DataFrames (e.g., `AgentSetDF.agents`, `Grid.cells`, etc.). While useful, this approach lacks flexibility when users need to access specific subsets of data.

Proposed Enhancement:
Implement indexable properties that support masking, allowing more granular data access. This would enable syntax like `AgentSetDF.agents[ids]` or `Grid.cells[coords]`.

Implementation Details:
1. Utilize the [descriptor protocol](https://docs.python.org/3/howto/descriptor.html) to maintain the current behavior when accessing the full property, while adding support for indexing and masking.
2. Ensure that the `__getitem__` method of the descriptor returns a DataFrame subset when a mask is provided.
3. Implement a setter method to allow operations like `Grid.cells[coords] = properties`, enhancing API usability.

Benefits:
- More intuitive and flexible data access
- Improved code readability and efficiency when working with subsets of data
- Consistent API across different components of the library

Example Usage:
```python
# Current: Returns full DataFrame
all_agents = agent_set.agents

# Proposed: Returns subset of agents
specific_agents = agent_set.agents[specific_ids]

# Proposed: Set properties for specific cells
grid.cells[specific_coords] = new_properties
```

Start by locating the implementations of AgentSetDF.agents and Grid.cells, then review Python's descriptor protocol documentation. Define how full-property access, masking through __getitem__, and assignment through the setter should behave for both examples. Done means indexed reads return the intended DataFrame subsets and indexed writes update selected properties without breaking existing full-property access.

Written by the indexing model from the issue text.