fix: unskip evaluator integ test classes in sm-train

[lucasjia-aws]

Issue

Some of the integration tests were passed, working on unskipping them to improve test cover rate.

Description

This PR unskips evaluator integration test classes in sagemaker-train that were temporarily skipped during a file relocation and never re-enabled.

Changes Summary

#	Test	Module	Action	Commit	How Fixed
1	TestLLMAsJudgeEvaluatorIntegration (class)	test_llm_as_judge_evaluator.py	Unskipped	12779438	Tests pass as-is
2	TestBenchmarkEvaluatorIntegration (class)	test_benchmark_evaluator.py	Unskipped	12779438	Tests pass as-is
3	TestCustomScorerEvaluatorIntegration (class)	test_custom_scorer_evaluator.py	Unskipped	12779438	tests pass as-is
4	test_llm_as_judge_builtin_metrics_only	test_llm_as_judge_evaluator.py	Deleted	f58d8368	Redundant with full_flow; replaced by unit tests for None metrics handling
5	test_llm_as_judge_custom_metrics_only	test_llm_as_judge_evaluator.py	Deleted	f58d8368	Redundant with full_flow; replaced by unit tests for None metrics handling
6	test_custom_scorer_with_builtin_metric	test_custom_scorer_evaluator.py	Unskipped	107e5fbe	Test passes as-is
7	test_custom_scorer_base_model_only	test_custom_scorer_evaluator.py	Unskipped	107e5fbe	Passes (test body is pass — placeholder for future implementation)
8	test_benchmark_evaluation_base_model_only	test_benchmark_evaluator.py	Unskipped + Fixed	3ce1324c, ab553e6c	Updated config from old account to test account; commented out non-existent mlflow_resource_arn
9	test_benchmark_evaluation_nova_model	test_benchmark_evaluator.py	Unskipped (still failing)	107e5fbe	Requires us-east-1 infra in test account — needs separate infra work
10	TestCustomScorerEvaluatorIntegration	test_custom_scorer_evaluator.py	Marked serial	b1c094e0	Added @pytest.mark.xdist_group to prevent pipeline concurrency conflicts

Test Results

• Passing: All evaluator tests except test_benchmark_evaluation_nova_model and test_benchmark_evaluation_base_model_only
• Flaky: test_custom_scorer_evaluation_full_flow (intermittent pipeline concurrency issue)
• Still failing (mark back as skip to merge this pr):
  • test_benchmark_evaluation_nova_model — requires cross-account us-east-1 infrastructure
  • test_benchmark_evaluation_base_model_only — pipeline creation returns None (under investigation)

New Unit Tests Added

• test_llm_as_judge_evaluator_builtin_metrics_only_no_custom — verifies evaluator handles custom_metrics=None correctly
• test_llm_as_judge_evaluator_custom_metrics_only_no_builtin — verifies evaluator handles builtin_metrics=None correctly

[lucasjia-aws]

succeeded test rerun:

• serve: https://us-west-2.console.aws.amazon.com/codesuite/codebuild/729646638167/projects/sagemaker-python-sdk-ci-sagemaker-serve-integ-tests/build/sagemaker-python-sdk-ci-sagemaker-serve-integ-tests%3Ab975a68e-2e99-4816-b330-e70bebf90d7d?region=us-west-2
• mlops: https://us-west-2.console.aws.amazon.com/codesuite/codebuild/729646638167/projects/sagemaker-python-sdk-ci-sagemaker-mlops-integ-tests/build/sagemaker-python-sdk-ci-sagemaker-mlops-integ-tests%3A62df11f0-3d65-4244-a406-9e0bdfc3cb55?region=us-west-2