feat: add MiniMax image generation tool

[octo-patch]

Reason: The built-in image-generation tool registry has no MiniMax implementation or request/response schema.

This change:

• adds a MiniMax text-to-image tool with a region-configurable endpoint, validated generation arguments, and URL/base64 response parsing
• registers provider and tool metadata, credential masking, and UI configuration
• covers payload construction and response parsing with focused unit tests

Checks:

• uvx ruff check src/backend/bisheng_langchain/gpts/tools/api_tools/minimax_image.py src/backend/bisheng_langchain/gpts/tools/api_tools/minimax_image_core.py src/backend/test/tool/test_minimax_image.py
• python3.11 -m unittest src.backend.test.tool.test_minimax_image
• jq empty on the updated tool registries and locale files
• python3.11 -m py_compile on the new provider modules
• git diff --check