Consistent data curation scripts location

[cwing-nvidia]

Use cases, pain points, and background
Today there’s no single, consistent place or name for “the data script” for each resource server.

Data scripts are generally located in each resource server, but they are named inconsistently. Most are at the root of the resource server; only reasoning_gym uses a scripts/ subfolder.

Resource server	Where the script lives	Script name(s)
newton_bench	Root (next to app.py)	generate_dataset.py
reasoning_gym	scripts/	scripts/create_dataset.py
ns_tools	Root	prepare_dataset.py
math_formal_lean	Root	prepare_nemotron_math_proofs.py, prepare_minif2f.py
multichallenge	Root	dataset_preprocess.py
arc_agi	Root	create_dataset.py

Description:
Recommend to standardize data scripts in resources_servers/<name>/data_curation/