docs: add dark-mode audio recognition architecture diagram

[icecold009]

Summary

• add a dark-mode GitDiagram PNG generated from the public repository default branch
• add Mermaid source covering CLI/web input, the shared audio contract, preprocessing, deterministic local fingerprint matching, optional providers, result reporting, and all public terminal statuses
• link the diagram and source review paths from the README

Evidence

• GitDiagram page: https://gitdiagram.com/icecold009/Audio-Recognition
• repository links checked against the public main branch for main.py, Flask routes, recorder/audio normalization, fingerprinting, matcher/provider adapters, configuration, and tests
• Markdown-relative-link check: 57 targets passed
• architecture-content check passed
• .venv-pipeline\\Scripts\\python.exe -m pytest -q: 193 passed
• PNG visually rechecked in dark mode at 1773x1279

Evidence boundaries

The diagram does not claim recognition accuracy, active provider credentials, hosted deployment, browser/device capture quality, or a completed lawful benchmark. Provider adapters remain optional/configuration-dependent, and the local constellation-hash path is described as deterministic implementation behavior rather than measured accuracy.

No deployment or merge is included in this PR.