Add IBM Watson transcription provider

[chinook1001]

Summary

• add an ibm_watson provider for IBM Watson Speech to Text using the synchronous /v1/recognize REST API
• read credentials from IBM_WATSON_STT_API_KEY and IBM_WATSON_STT_URL, with an optional IBM_WATSON_STT_TIMEOUT
• add mocked tests for availability gates, request shape, content type mapping, response parsing, error handling, and model aliases

Validation

• PYTHONPATH=/Users/lucas/Documents/codex-bounty-work/sapat-ibm-watson /tmp/sapat-eleven-venv/bin/python -m pytest tests/providers/test_ibm_watson.py tests/test_registry.py -q -> 18 passed, 1 local LibreSSL/urllib3 warning
• PYTHONPATH=/Users/lucas/Documents/codex-bounty-work/sapat-ibm-watson /tmp/sapat-eleven-venv/bin/python -m black --check sapat/providers/ibm_watson.py tests/providers/test_ibm_watson.py
• PYTHONPATH=/Users/lucas/Documents/codex-bounty-work/sapat-ibm-watson /tmp/sapat-eleven-venv/bin/python -m compileall sapat/providers/ibm_watson.py tests/providers/test_ibm_watson.py
• git diff --check

Companion implementation for the Daytona content bounty.

No secrets, audio files, transcripts, payout details, or private account data are included.

[chinook1001]

Validation update for current head d155dc4deca1bdb20547446eb576f0d7a6ffa2ea:

• PYTHONPATH=/Users/lucas/Documents/codex-bounty-work/sapat-ibm-watson /tmp/sapat-verify-venv/bin/python -m pytest tests -q -> 184 passed, 1 local LibreSSL/urllib3 warning.
• PYTHONPATH=/Users/lucas/Documents/codex-bounty-work/sapat-ibm-watson /tmp/sapat-verify-venv/bin/python -m pytest tests/providers/test_ibm_watson.py tests/test_registry.py -q -> 18 passed, 1 local LibreSSL/urllib3 warning.
• PYTHONPATH=/Users/lucas/Documents/codex-bounty-work/sapat-ibm-watson /tmp/sapat-verify-venv/bin/python -m black --check sapat/providers/ibm_watson.py tests/providers/test_ibm_watson.py -> pass.
• git diff --check origin/main...HEAD -> pass.

The combined GitHub status is currently red because the external Continuous AI agent statuses report agent errors; I do not see a failing repository test/check-run attached to this head. No secrets or live IBM credentials were used; provider behavior is covered with mocked HTTP calls.

[chinook1001]

Note on current failing GitHub status: the failing contexts are from the external Continuous AI agents reporting Agent encountered an error, not from the project test suite. I rechecked the branch locally and posted the validation commands/results above; happy to address any maintainer-visible code/test failure if one appears.