Add STT.ai transcription provider

[abridegan-bit]

Summary

• add an STT.ai transcription provider exposed as --api sttai
• support endpoint/model/diarization/speaker/response-format settings through STTAI_* environment variables
• document STT.ai configuration and add mocked unit coverage for request construction, anonymous mode, and missing-file validation

Validation

• /tmp/sapat-venv/bin/python -m unittest discover -s tests -v
• /tmp/sapat-venv/bin/python -m compileall src tests
• /tmp/sapat-venv/bin/sapat --help
• git diff --check

Related Daytona bounty issue: daytonaio/content#13

[nibzard]

Thanks for the contribution, but after reviewing all incoming PRs we've decided not to merge this one.

Reason: STT.ai is not a well-documented or established STT provider, and the implementation allows anonymous requests with no API key, raising questions about whether this is a production API.

We've refactored the codebase into a provider plugin architecture (see PR coming soon on feat/provider-plugin-architecture) and are incorporating provider support differently. Closing this PR.