Feat: Row-level eval details

[JeanKaddour]

This pull request implements row-level eval details. The changes involve modifications to the evaluation logic, updates to the frontend components, and the addition of a new results page.

Backend Changes:

1. Evaluation Logic Updates:
   • Replaced task_id with example_id in async def evaluate_dataset_batch to better reflect the context of the evaluations. [1] [2] [3]
   • Added full_prompt to the loop in async def evaluate_dataset_batch to include the full prompt in the evaluation.
   • Updated the response storage and logging to use example_id instead of task_id.

Frontend Changes:

1. Component Cleanup:

   • Removed unused imports from Header.jsx to clean up the code.

2. Evaluation Page Updates:

   • Simplified the handleViewResults function in evals.jsx to navigate to a dedicated results page instead of using a modal.
   • Removed the modal for displaying evaluation results from evals.jsx.

3. New Results Page:

   • Added a new results page (evals/[id].js) to display evaluation results in a table format, including example ID, problem, predicted answer, ground truth, and correctness status. (frontend/src/pages/evals/[id].jsR1-R164)