Fix lm_eval.tasks' has no attribute 'initialize_tasks' error

[KeitaW]

What does this PR do?

Fixes #422

lm-evaluation-harness has introduced breaking changes causing lm_eval.tasks' has no attribute 'initialize_tasks' error in the current recipes/evaluation/eval.py. This PR is aiming to resolve the issue by adopting new TaskManager API.
cf. https://github.com/EleutherAI/lm-evaluation-harness/releases

Feature/Issue validation/testing

Please describe the tests that you ran to verify your changes and relevant result summary. Provide instructions so it can be reproduced.
Please also list any relevant details for your test configuration.

python/eval.py \
     --model hf --model_args pretrained=meta-llama/Meta-Llama-3-8B --tasks hellaswag --device cuda:0   --batch_size 8"

yields

Map: 100%|██████████| 10042/10042 [00:01<00:00, 8628.17 examples/s]
0: 2024-05-06:12:44:13,769 INFO     [task.py:395] Building contexts for hellaswag on rank 0...
 62%|██████▏   | 6276/10042 [00:01<00:01, 2603.76it/s]
100%|██████████| 10042/10042 [00:02<00:00, 3578.18it/s]
0: 2024-05-06:12:44:17,334 INFO     [evaluator.py:393] Running loglikelihood requests
...
0: 2024-05-06:12:47:36,076 INFO     [eval.py:39] {
0:   "results": {
0:     "hellaswag": {
0:       "acc,none": 0.6015733917546305,
0:       "acc_stderr,none": 0.004885735963346892,
0:       "acc_norm,none": 0.7905795658235412,
0:       "acc_norm_stderr,none": 0.004060633907027273,
0:       "alias": "hellaswag"
0:     }
0:   },
...

Before submitting

• [] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you read the contributor guideline,
  Pull Request section?
• Was this discussed/approved via a Github issue? Please add a link
  to it if that's the case.
• Did you make sure to update the documentation with your changes?
• Did you write any new necessary tests?

Thanks for contributing 🎉!

[HamidShojanazeri]

Thanks @KeitaW for the PR.