feat: add MiniMax as third LLM provider

[octo-patch]

Summary

Add MiniMax AI as a first-class LLM provider alongside OpenAI and Azure OpenAI. MiniMax offers an OpenAI-compatible API with models like MiniMax-M2.7 (1M context window) and MiniMax-M2.7-highspeed.

Changes

Backend

• Add MINIMAX enum value to ModelType and minimaxApiKey field to KeyConfiguration interface
• Add MINIMAX_API_KEY / MINIMAX_API_MODEL environment variables in const.ts
• Handle MiniMax configuration in configuration.ts (headers + env vars + validation)
• Route MiniMax chat models via OpenAI-compatible basePath (https://api.minimax.io/v1) in openai.ts
• Add MiniMax embedding support using embo-01 model in embeddings.ts

Frontend

• Add MiniMax tab in KeySettings.tsx with API key input and model selector (M2.7 / M2.7-highspeed)
• Pass x-minimax-api-key header in all API calls (chat, query, embedding)
• Update validation to recognize MiniMax API key
• Rename settings button/title from "OpenAI" to generic "API Key Settings"

Documentation

• Add MiniMax to feature list in README.md
• Add MINIMAX_API_KEY and MINIMAX_API_MODEL to doc/env-vars.md

Tests (31 total)

• Unit tests: types.test.ts (4), configuration.test.ts (8), openai.test.ts (7), embeddings.test.ts (4)
• Integration tests: integration.test.ts (5) — end-to-end config → model/embeddings flow
• All tests passing

How to use

Via environment variables

OPENAI_TYPE=MINIMAX
MINIMAX_API_KEY=your-minimax-api-key
MINIMAX_API_MODEL=MiniMax-M2.7

Via UI

Click API Key Settings → select the MiniMax tab → enter your API key and choose a model.

Test plan

• All 31 unit and integration tests pass (npx jest)
• Manual verification: select MiniMax provider in UI, enter API key, send a chat message
• Manual verification: upload a file and chat with it using MiniMax provider