feat: upgrade MiniMax provider preset and add model presets

[octo-patch]

Summary

• Fix MiniMax API base URL: Updated from the deprecated api.minimax.chat to the correct api.minimax.io endpoint
• Add MiniMax model presets: Added M2.7, M2.7-highspeed (1M context), M2.5, M2.5-highspeed (204K context) to MODEL_PRESETS so users can quick-add MiniMax models when configuring the provider
• Update description: Refreshed MiniMax provider description to mention the latest M2.7/M2.5 model series instead of the outdated "MiniMax-Text" series
• README updates: Added MiniMax to the provider tables in both Chinese and English READMEs
• Unit tests: Added 12 unit tests covering MiniMax provider preset validation, model preset structure, context window sizes, and provider-model alignment

Why

MiniMax has migrated their API from api.minimax.chat to api.minimax.io and released new M2.7 models with 1M context window. The existing preset pointed to the old endpoint and had no model presets, making it harder for users to get started with MiniMax.

Test plan

• All 12 new tests pass: node --test tests/model-presets.test.js
• Existing tests still pass: node --test tests/gateway-guardian-policy.test.js
• Manual: select MiniMax from provider presets - verify base URL is https://api.minimax.io/v1
• Manual: click Add Model on MiniMax provider - verify M2.7/M2.5 quick-add buttons appear