Fix: default to zero input when none provided in integrate

[ftavella]

Summary

Fixes #18

When input is not specified in integrate (or __call__), the model now defaults to an array of zeros rather than raising a ValueError. This makes input truly optional, consistent with how initial_condition already works.

Changes

• integrate: replace raise ValueError when input is None with a default of np.zeros(len(time)) for single-input models and np.zeros((len(time), num_inputs)) for multi-input models
• Also fixes pkg_resources import in setup.py for setuptools >= 82 compatibility

Tests

Updated test_models.ipynb:

• Removed the test that expected a ValueError when no input is provided
• Added test verifying that calling a single-input model (Forger99) without input produces the same trajectory as passing np.zeros
• Added test verifying that a multi-input model runs without error when input is omitted