Add NAdam optimizer with unit tests

[arunpandianj]

Summary

• Added NAdam optimizer (Nesterov-accelerated Adam) to Qiskit Machine Learning.
• Implemented minimize, snapshot saving/loading, callback support.

Details

• Added nadam.py to optimizers.
• Updated __init__.py to include NAdam.
• Added unit tests in test/optimizers/test_nadam.py.
• Tests verify: basic optimization, noisy optimization, callback, settings, snapshot save/load.

Checks

• Unit tests pass
• Follows Qiskit code style

[CLAassistant]

All committers have signed the CLA.

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[arunpandianj]

This is a Python implementation of the NAdam optimizer for Qiskit, combining Adam’s adaptive moment estimation with Nesterov momentum for faster convergence. It supports numerical gradient computation, optional gradient noise, and bias-corrected moment updates. The optimizer includes features like iteration snapshots, callback functions, and convergence checking via a tolerance threshold. While it currently ignores analytical gradients, it can save and reload internal states (_m, _v, _t) from snapshots, making it suitable for long-running optimization tasks.