Add evaluation test scenarios and remove orphaned azure-ai-evaluation-py tests

[imatiach-msft]

Summary

Adds comprehensive evaluation test scenarios to azure-ai-projects-py and removes the orphaned azure-ai-evaluation-py test folder.

Changes

Added to tests/scenarios/azure-ai-projects-py/scenarios.yaml

6 new evaluation scenarios (+514 lines):

Scenario	Description
evaluation_with_inline_data	Complete eval workflow: inline JSONL data, create eval, run, poll, retrieve results
agent_evaluation_with_sample	Agent evaluation with {{sample.output_text}} and {{sample.output_items}} mapping
custom_code_evaluator	CodeBasedEvaluatorDefinition with project_client.evaluators.create_version()
safety_evaluators	builtin.violence, builtin.sexual, builtin.hate_unfairness
openai_graders	label_model, string_check, text_similarity grader types
list_builtin_evaluators	Evaluator discovery via project_client.evaluators.list_latest_versions()

Deleted

• tests/scenarios/azure-ai-evaluation-py/ - orphaned after skill was merged into azure-ai-projects-py in PR fix: Skills grid cropping and documentation cards consistency #93

Test Results

All 18 scenarios pass with 100% score:

Scenarios: 18 Passed: 18 Failed: 0 Pass Rate: 100.0% Average Score: 100.0

Notes

The new scenarios ensure the deprecated azure-ai-evaluation SDK patterns are flagged as forbidden:

• from azure.ai.evaluation import → forbidden
• ViolenceEvaluator, @evaluator decorator → forbidden
• Class-based graders like AzureOpenAILabelGrader → forbidden