Add DeepInfra transcription provider

[justusaugust]

Summary

• add --api deepinfra backed by DeepInfra native speech recognition
• document DeepInfra environment variables and optional chat correction config
• add mocked unit coverage for config validation, request construction, endpoint override, API errors, and CLI routing

Verification

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

Companion content PR for the Daytona guide will link back here.

[justusaugust]

Local validation for the DeepInfra provider is passing:

• .venv/bin/python -m unittest discover -s tests -v (6 tests OK)
• .venv/bin/python -m compileall src tests
• .venv/bin/sapat --help | grep deepinfra

The three red Continuous AI statuses currently report Agent encountered an error and do not include code feedback or a failing test log in GitHub. No secrets or media files are committed.