fix: Resolves HumanEval Inference by replacing task_id with instance_id

[Ethan0456]

End-user friendly description of the problem this fixes or functionality that this introduces

This PR resolves an issue in the HumanEval benchmark's run_infer.py script where the code incorrectly references a column task_id that no longer exists. The correct column is instance_id. This fix ensures that the script runs without errors, making it compatible with the latest schema changes.

• Include this change in the Release Notes.
  End-user friendly description for the Release Notes:
  "Fixes a column reference issue in run_infer.py where task_id was incorrectly used instead of instance_id."

---

Give a summary of what the PR does, explaining any non-trivial design decisions

This PR refactors the Humaneval benchmark's run_infer.py script by replacing all occurrences of task_id with instance_id to align with the updated column naming convention. No major design changes were made, but the code now references the correct column to avoid attribute errors and ensure smooth execution of the Humaneval benchmark.

---

Link of any specific issues this addresses

This PR addresses issue #4363, where an AttributeError occurs because the column name change from task_id to instance_id was not consistently applied throughout the file, leading to the error.

[xingyaoww]

Thanks for catching the bug!