feat: OpenRouter LLM integration (#95)

[faisalnugroho]

Summary

Implements OpenRouter integration for the Lux framework, providing access to 200+ LLM models through a single unified API.

Features

• OpenRouter API Client: Full implementation compatible with OpenAI chat completions format
• Model Discovery: Browse available models with filtering
• Cost Estimation: Budget planning with get_model_cost and estimate_cost
• Provider Preferences: Route requests to specific providers
• Fallback Support: Automatic failover to alternative models
• Cost Tracking: Built-in telemetry for monitoring API costs
• Error Handling: Proper handling of rate limits, auth errors, and transport failures

Acceptance Criteria

• Implement OpenRouter API client
• Add support for multiple models
• Create unified interface matching other providers
• Implement proper error handling and retries
• Add cost tracking and reporting
• Include comprehensive documentation
• Add integration tests

Usage

Basic usage with OpenRouter:

Lux.LLM.OpenRouter.call("Hello!", [], %{model: "openai/gpt-4o"})

With provider preferences:

Lux.LLM.OpenRouter.call("Hello!", [], %{
  model: "anthropic/claude-3.5-sonnet",
  provider_preferences: %{allow_fallbacks: true}
})

List available models:

{:ok, models} = Lux.LLM.OpenRouter.list_models(filter: "anthropic")

Estimate cost:

cost = Lux.LLM.OpenRouter.estimate_cost("openai/gpt-4o", 1000, 500)

Closes #95

[faisalnugroho]

Hi! Just following up on this PR. Let me know if there are any changes needed or if you have any questions about the implementation.

[faisalnugroho]

Hi! Just following up on this PR. OpenRouter LLM integration. All checks pass — would appreciate your review when you get a chance. Thanks!