Fixing LitQATaskDatasetcompute_trajectory_metrics crash with bad status extraction

[jamesbraza]

If the re.split(pattern=GenerateAnswer.ANSWER_SPLIT_REGEX_PATTERN, ...) doesn't give at least 4 values, the ensuing total paper count, relevant paper count, and evidence count summation will crash with an IndexError:

  File "/path/to/.venv/lib/python3.12/site-packages/paperqa/agents/task.py", line 193, in compute_trajectory_metrics
    sum(int(sa[i]) for sa in split_answers) / len(split_answers)
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/path/to/.venv/lib/python3.12/site-packages/paperqa/agents/task.py", line 193, in <genexpr>
    sum(int(sa[i]) for sa in split_answers) / len(split_answers)
            ~~^^^
IndexError: list index out of range

This PR fixes that IndexError