feat(cli): add options to example runner

[meiiie]

Motivation

The quickstart script works well for the default example, but changing common runtime settings currently requires editing run_misotts.py. This adds CLI options for those settings while keeping the default example path unchanged.

Changes

• Add flags for model source, device, output path, max audio length, temperature, and top-k.
• Keep python run_misotts.py behavior unchanged: auto-select CUDA when available, use MISO_TTS_8B_MODEL or MisoLabs/MisoTTS, and write full_conversation.wav.
• Move heavyweight imports behind argument parsing so --help does not initialize the inference stack.
• Add stdlib unittest coverage for parser defaults, overrides, and invalid generation settings.

Validation

• py -3.10 -m unittest
• py -3.10 run_misotts.py --help
• py -3.10 -m py_compile run_misotts.py tests/test_run_misotts.py
• git diff --check origin/main...HEAD

Not run

• Full TTS inference; this local machine does not have enough CUDA VRAM for the 8B checkpoint.

[AodenTeo]

I definitely like this change. I'll review it later.

[meiiie]

I definitely like this change. I'll review it later.

Thank you, I would be happy to contribute to the project. Also, I have a question: will MisoTTS be expanded to other languages?

[devin-ai-integration]

✅ Devin Review: No Issues Found

Devin Review analyzed this PR and found no bugs or issues to report.