Rewrite azure-ai-evaluation skill to use azure-ai-projects SDK (V2 API)

[imatiach-msft]

Summary

This PR deprecates the azure-ai-evaluation-py skill and merges its functionality into azure-ai-projects-py using the current V2 API.

Background

Per guidance from April:

"azure-ai-evaluation is v1 API (not compatible with nextgen and eventually will be deprecated)" "don't use or reference azure-ai-evaluation anywhere"

Changes

Deleted

• .github/skills/azure-ai-evaluation-py/ - Entire folder deleted (deprecated SDK)

Enhanced azure-ai-projects-py skill

• references/evaluation.md - Completely rewritten with comprehensive evaluation patterns
• references/built-in-evaluators.md - New file with complete evaluator reference (Quality, Safety, Agent, NLP)
• references/custom-evaluators.md - New file with code-based and prompt-based evaluator patterns
• scripts/run_batch_evaluation.py - New CLI tool for batch evaluations
• SKILL.md - Updated reference files list, renamed product to "Microsoft Foundry"

Key API Pattern (V2)

   from azure.ai.projects import AIProjectClient

   with AIProjectClient(endpoint=endpoint, credential=credential) as project_client:
       openai_client = project_client.get_openai_client()

       eval_object = openai_client.evals.create(
           name="My Evaluation",
           testing_criteria=[{
               "type": "azure_ai_evaluator",
               "evaluator_name": "builtin.coherence",
               "data_mapping": {"query": "{{item.query}}", "response": "{{item.response}}"}
           }]
       )
       run = openai_client.evals.runs.create(eval_id=eval_object.id, data_source=data_source)

References

• Azure AI Projects evaluation samples (https://github.com/Azure/azure-sdk-for-python/tree/main/sdk/ai/azure-ai-projects/samples/evaluations)

[thegovind]

LGTM