feat: polish audio recognition showcase experience

[icecold009]

Summary

• ship the polished Flask showcase UI and responsive typography
• replace the microphone control with the outlined icon treatment
• align README, product, technical, UI, backend, and showcase materials with the current implementation boundary
• keep provider, benchmark, auth, persistent history, and public deployment claims explicit

Verification

• 193 tests passed
• 82% source-scoped branch coverage
• Ruff, formatting, compilation, and diff checks passed
• browser checks passed at desktop, 390px, and 320px with no console errors or horizontal overflow
• CI run: https://github.com/icecold009/Audio-Recognition/actions/runs/34850459846

Known limitations

• provider credentials and credentialed live recognition remain unverified
• complete lawful real-world benchmark corpus remains unavailable
• authentication, persistent history, and public deployment are not implemented product features

This PR is intentionally scoped to the showcase-ready local MVP and its documentation.