Add example: multi-provider comparison (same task, different LLMs)

[rajkumar42]

Summary

Create an example that runs the same agent task across multiple LLM providers (Claude, OpenAI, Gemini) and compares the outputs side-by-side. Demonstrates provider-agnostic design and helps users evaluate which provider works best for their use case.

What needs to happen

• Create multi_provider_comparison/ directory
• Define a single task (e.g., "summarize this article" or "extract key entities")
• Run the task against Claude, OpenAI, and Gemini
• Collect and display results side-by-side (output, tokens, latency, cost)
• Add README explaining setup (API keys) and how to run

Example output

Task: "Summarize the key points of this earnings report"

| Provider | Tokens (in/out) | Latency | Output preview           |
|----------|----------------|---------|--------------------------|
| Claude   | 1200/350       | 2.1s    | "Revenue grew 15%..."    |
| OpenAI   | 1180/420       | 1.8s    | "The company reported..." |
| Gemini   | 1250/380       | 2.4s    | "Key highlights..."      |

Acceptance criteria

• Runs against at least 2 providers (3 if all API keys available)
• Output comparison is clear and useful
• Gracefully handles missing API keys (skips that provider with a message)