feat(engine, nn): add tanh/exp/log/sigmoid ops, flexible activations, and unittest suite

[miskatul-anwar]

Summary

This PR expands the autograd capabilities of micrograd and introduces a complete, standalone unit test suite:

• Autograd Engine (micrograd/engine.py): Added tanh(), exp(), log(), and sigmoid() methods to the Value class with analytical gradient backpropagation.
• Neural Networks (micrograd/nn.py): Enhanced Neuron to support 'tanh', 'sigmoid', and linear activations alongside 'relu'.
• Test Suite (test/test_engine.py, test/test_nn.py): Refactored tests to Python's standard library unittest framework with graceful fallbacks when PyTorch is not installed, and added a dedicated test suite covering Neuron, Layer, and MLP modules.

Testing & Verification

Ran the full test suite via python3 -m unittest discover -s test:

..........
----------------------------------------------------------------------
Ran 10 tests in 0.001s

OK