Add input validation for model features shape

[Koolvansh07]

Pull Request

Description

This PR addresses issue #190 by adding explicit input validation to the forward methods of GraphWeatherForecaster, GraphWeatherAssimilator, and AuroraModel.

The models now strictly enforce that input feature tensors must be 3-dimensional with the shape [batch, nodes, features]. Previously, invalid input shapes (e.g., missing the feature dimension) could pass through to internal layers, causing cryptic runtime errors deep in the call stack. This change ensures that such errors are caught immediately at the API boundary with a clear and informative ValueError.

Fixes #190

How Has This Been Tested?

I have added a new test file tests/test_input_validation.py which contains unit tests for each of the modified models.

To reproduce/verify:

1. Run pytest tests/test_input_validation.py

The tests verify that:

• Passing an input tensor with shape [batch, nodes] (2D) raises a ValueError with the expected message.

• Passing an input tensor with shape [batch, nodes, features] (3D) does not raise a shape-related error.

• Yes

If your changes affect data processing, have you plotted any changes? i.e. have you done a quick sanity check?

• Yes (N/A - this is a validation fix, not a data processing change)

Checklist:

• My code follows OCF's coding style guidelines
• I have performed a self-review of my own code
• I have made corresponding changes to the documentation (N/A - internal validation logic)
• I have added tests that prove my fix is effective or that my feature works
• I have checked my code and corrected any misspellings