Add llama 3.2 mmlu, math, gpqa evals to meta_eval harness (#801)

tools/benchmarks/llm_eval_harness/meta_eval/README.md

tools/benchmarks/llm_eval_harness/meta_eval/eval_config.yaml

...eval_harness/meta_eval/meta_template/gpqa/gpqa_0shot.yaml

...ks/llm_eval_harness/meta_eval/meta_template/gpqa/utils.py

...ness/meta_eval/meta_template/gpqa_cot/gpqa_0shot_cot.yaml

...ess/meta_eval/meta_template/math_hard/math_4shot_cot.yaml

...eta_eval/meta_template/math_hard/math_hard_0shot_cot.yaml

...s/llm_eval_harness/meta_eval/meta_template/mmlu/mmlu.yaml

...ks/llm_eval_harness/meta_eval/meta_template/mmlu/utils.py

...enchmarks/llm_eval_harness/meta_eval/prepare_meta_eval.py