feat: add MiniMax as alternative backend provider

[octo-patch]

Summary

• Add MiniMax as an alternative backend provider alongside Azure OpenAI
• MiniMax provides an OpenAI-compatible API, so the proxy can seamlessly route requests to either Azure or MiniMax based on provider_type configuration
• Support for MiniMax-M2.7, MiniMax-M2.5, and MiniMax-M2.5-highspeed models

Changes

• azure/model.go: Add ProviderType field to DeploymentConfig; add MiniMaxConverter implementing RequestConverter for MiniMax API routing (Bearer auth, no api-version param)
• azure/init.go: Add InitMiniMaxFromEnv() for MINIMAX_API_KEY/MINIMAX_ENDPOINT/MINIMAX_MODEL_MAPPER environment variable support with default model registration
• azure/proxy.go: Auto-select converter based on provider_type; update ModelProxy() to return static MiniMax model info alongside Azure deployments
• constant/env.go: Add MiniMax environment variable constants
• README.md: Document MiniMax configuration (env vars, Docker, config file, mixed Azure+MiniMax setup)

Test Plan

• 9 unit tests: converter name, path preservation, auth handling, streaming, custom endpoint, deployment config, backward compatibility
• 3 integration tests with mock HTTP servers: chat completions proxy, model list, streaming proxy
• All 12 tests pass
• Verify backward compatibility: existing Azure-only configurations work unchanged