add agent evaluation example with @observe and component-level tracing

[Ajay6601]

Adds examples/agent_evaluation/ with a complete, runnable example showing DeepEval v3.0 agent evaluation capabilities:

test_agent_eval.py
Three ways to evaluate an agent:

1. evaluate () function (quickest)
2. pytest integration (deepeval test run)
3. Component-level eval with @observe and update_current_span
   README.md with Quick start guide with metric descriptions

Uses a mock agent with retriever to keep the example (no external API calls needed beyond the evaluation LLM).

[vercel]

@Ajay6601 is attempting to deploy a commit to the Confident AI Team on Vercel.

A member of the Team first needs to authorize it.

[penguine-ip]

Hey @Ajay6601 the observed callback is no longer supported - it is the evals iterator right now: https://deepeval.com/docs/evaluation-component-level-llm-evals#run-component-level-evals

[Ajay6601]

Thanks for the review! Updated to use evals_iterator with @observe + update_current_span per the current docs. The example now demonstrates trace-level metrics via evals_iterator(metrics=[ ]) and span-level metrics via @observe(metrics=[ ]), with proper nested spans forming the trace hierarchy.