[codex] Add OpenAI voice field test with automatic log upload

[Anthonyiswhy]

Summary

• Adds the v3.30 OpenAI voice field-test branch with optional OpenAI alert TTS and push-to-talk command input.
• Adds optional GitHub upload for completed run logs via BLINDNAV_LOG_UPLOAD=1.
• Adds log retention so ~/blindnav_logs keeps only the newest navigation runs by default, plus startup recovery upload for a previous completed run if a prior shutdown failed.
• Makes the TTS comparison helper scripts executable.

Validation

• python -m pytest tests/test_blindnav.py tests/test_blindnav_v326.py -q -> 195 passed on rerun.
• python -m py_compile raspberry_pi/yolo_realsense_navigation.py tools/upload_run_logs_to_github.py

Notes

• Default log retention is BLINDNAV_LOG_RETENTION_RUNS=10 navigation runs.
• Startup recovery upload only runs when BLINDNAV_LOG_UPLOAD=1.

[Anthonyiswhy]

Added docs/HANDOFF_V330_LOG_UPLOAD.md to the PR branch with field-test commands, log-upload behavior, validation notes, and the current recommendation not to replace OpenAI STT with JAX Whisper unless it benchmarks well under live Pi load.