fix(evaluation): link traced runs to examples - #1

closes: langchain-ai#208

Summary

• Always link traced evaluation runs to their source dataset example.
• Add regression coverage for successful evaluations using EvaluateErrorHandling.IGNORE.

Problem

referenceExampleId was set only when EvaluateErrorHandling.LOG was selected. Successful traced runs created with other error-handling modes were therefore not associated with their dataset example.

Testing

• git diff --check
• Focused EvaluateRunnerTest run was attempted but exceeded the local execution timeout.