Add MiniMax as alternative LLM provider for ChatOCR and DocTranslation

[octo-patch]

Summary

Add MiniMax Cloud as an alternative LLM provider for the PP-ChatOCRv4 and PP-DocTranslation pipelines, alongside the existing Qianfan (ERNIE) support.

What's Changed

• New llm_config.py module: Shared helper get_minimax_chat_bot_config() that builds a chat_bot_config dict for MiniMax's OpenAI-compatible API (MiniMax-M2.7, 204K context window)
• CLI integration: Added --minimax_api_key argument to both pp_chatocrv4_doc and pp_doctranslation CLI subcommands
• Environment variable support: API key can also be set via MINIMAX_API_KEY env var (more secure than CLI)
• Priority handling: When both --minimax_api_key and --qianfan_api_key are provided, MiniMax takes precedence
• Documentation: Updated both English and Chinese docs for PP-ChatOCRv4 and PP-DocTranslation with MiniMax usage examples (CLI and Python API)

Available MiniMax Models

Model	Context Window
MiniMax-M2.7 (default)	204K tokens
MiniMax-M2.7-highspeed	204K tokens

Usage

CLI:

Python API:

Files Changed (8 files, ~346 additions)

File	Change
paddleocr/_pipelines/llm_config.py	New: MiniMax config helper
paddleocr/_pipelines/pp_chatocrv4_doc.py	Add --minimax_api_key CLI arg
paddleocr/_pipelines/pp_doctranslation.py	Add --minimax_api_key CLI arg
tests/pipelines/test_minimax_llm_config.py	10 unit + 2 integration tests
docs/.../PP-ChatOCRv4.en.md	English docs update
docs/.../PP-ChatOCRv4.md	Chinese docs update
docs/.../PP-DocTranslation.en.md	English docs update
docs/.../PP-DocTranslation.md	Chinese docs update

Test Plan

• 10 unit tests pass (config structure, env var fallback, key precedence, error handling)
• 2 integration tests pass (MiniMax M2.7 and M2.7-highspeed API smoke tests)
• Existing pipeline tests unaffected (no changes to core pipeline logic)

[paddle-bot]

Thanks for your contribution!

[CLAassistant]

All committers have signed the CLA.

[luotao1]

Please fix the CodeStyle.

[changdazhou]

Thanks for your contribution. Please update the method to provide general support for the OpenAI API, rather than only supporting MiniMax.