Run experiment of MERA-evaluation

Generate docs on https://huggingface.co/datasets/MERA-evaluation/RealCode

Without public forks.

Snippet.
`from datasets import load_dataset
import pandas as pd

ds = load_dataset("MERA-evaluation/RealCode")
df = pd.DataFrame([{'repo': i['repo'],
'base_commit': i['base_commit']
} for i in ds['test']['meta']
]).drop_duplicates()`

Results should be presented in zip (example below):
[osa_gbfs_1208.zip](https://github.com/user-attachments/files/21776261/osa_gbfs_1208.zip)