during run evaluation of a test case inferences could be NoneType which has no len attribute and error is

## The problem

Reported in Google Agent Development Kit: during run evaluation of a test case inferences could be NoneType which has no len attribute and error is not caught.

## The fix

ADK eval crashes with len() on NoneType when a test case's inference failed: in local_eval_service.py the code calls len(inference_result.inferences) without a None check. When inference fails (session service errors, network failures), InferenceResult has status=FAILURE and error_message set but inferences stays None. The fix (PR #3805): in evaluate(), filter out failed inferences first, i.e. if inference_result.status == InferenceStatus.FAILURE or inference_result.inferences is None, log a warning and return EvalStatus.NOT_EVALUATED instead of evaluating. Upgrade ADK past the release containing that fix.