Add MiniMax as LLM translation and TTS provider

[octo-patch]

Summary

• Add MiniMax M2.5 / M2.7 as LLM translation providers (sequential & batch modes) via the OpenAI-compatible API at https://api.minimax.io/v1
• Add MiniMax TTS (speech-2.8-hd) with 12 verified voice presets as a new text-to-speech option
• Reuse existing gpt_sequential() / gpt_batch() by accepting an optional client parameter, minimizing code duplication
• Strip <think> tags and markdown code fences from MiniMax model responses before JSON parsing
• Add MINIMAX_API_KEY environment variable checks in both media and docs conversion flows

Changes

File	What
soni_translate/translate_segments.py	_create_minimax_client(), match-case routing for MiniMax models, think-tag & code-fence stripping
soni_translate/text_to_speech.py	segments_minimax_tts() function, regex pattern matching, 7-element return value
soni_translate/language_configuration.py	MINIMAX_TTS_MODELS list (12 voices)
app_rvc.py	Import TTS models, API key validation, Gradio dropdown integration
README.md	MiniMax setup instructions & credits
tests/	46 tests (18 unit translation + 19 unit TTS + 9 integration)

Test plan

• All 46 tests pass (python -m pytest tests/ -v)
• Set MINIMAX_API_KEY and run integration tests: python -m pytest tests/test_minimax_integration.py -v
• Select a MiniMax translation model in the Gradio UI and verify end-to-end translation
• Select a MiniMax TTS voice and verify audio generation

How to test locally

export MINIMAX_API_KEY='your-key'
python -m pytest tests/ -v                           # unit tests (no key needed)
python -m pytest tests/test_minimax_integration.py -v  # integration tests (key needed)