feat: add MiniMax as first-class LLM provider

[octo-patch]

Summary

Add MiniMax AI as a first-class LLM provider, using their OpenAI-compatible API at https://api.minimax.io/v1.

• Register minimax provider in llms.json with model aliases (large -> MiniMax-M2.7, small -> MiniMax-M2.7-highspeed) and pricing for M2.7/M2.5 model families
• Add MODEL_PROVIDER_MINIMAX constant and MINIMAX_API_BASE URL in constants.ts
• Add MINIMAX_API_KEY / MINIMAX_API_BASE environment variable parsing in env.ts
• Add configuration documentation page at docs/configuration/minimax.mdx
• Add unit tests (model parsing, provider registration, language model resolution)
• Add integration tests (env token parsing, custom base URL, missing key error)

How It Works

MiniMax provides an OpenAI-compatible chat completions API, so the existing LocalOpenAICompatibleModel handler works out of the box. Users configure via:

MINIMAX_API_KEY=eyJh...

Then use models like minimax:MiniMax-M2.7 or minimax:MiniMax-M2.5-highspeed.

Available Models

Model	Context	Description
MiniMax-M2.7	1M tokens	Latest flagship model
MiniMax-M2.7-highspeed	1M tokens	Faster variant
MiniMax-M2.5	204K tokens	Previous generation
MiniMax-M2.5-highspeed	204K tokens	Faster variant

Test Plan

• Unit tests: model identifier parsing for minimax models
• Unit tests: constants and base URL validation
• Unit tests: provider registered in defaultModelConfigurations aliases
• Unit tests: resolveLanguageModel returns OpenAI-compatible model
• Integration tests: parseTokenFromEnv with MINIMAX_API_KEY
• Integration tests: custom MINIMAX_API_BASE override
• Integration tests: error when MINIMAX_API_KEY is missing