[WIP] Add QML passive sonar reproduction

[jsenellart]

Summary

Adds a reproduction scaffold for Bach & Nguyen, “Quantum machine learning for passive sonar” under papers/qml_passive_sonar.

The reproduction includes:

• SAV spectral feature extraction and DEMON baseline
• CNN and HQ-CNN classifiers
• Pure-PyTorch statevector PQC simulator
• MerLin photonic HQ-CNN variant
• Synthetic fallback data for smoke runs
• On-demand DeepShip public-sample download when local WAVs are absent
• Data scaffold and dataset acquisition notes under data/qml_passive_sonar/
• Additional implementation/scaling notes in ADDITIONAL_NOTES.md
• Tests and plotting utilities

Data Policy

No audio datasets are committed. The repo tracks only:

• data/qml_passive_sonar/README.md
• data/qml_passive_sonar/deepship/.gitkeep
• data/qml_passive_sonar/shipear/.gitkeep

The data README documents:

• ShipsEar and DeepShip access links/contact information
• expected local layout
• class mappings
• DeepShip sample fallback behavior

Validation

Run from papers/qml_passive_sonar:

ruff check .
pytest -q

Results:

• ruff check .: passed
• pytest -q: 8 passed