fix: Upload other splits to Huggingface

[fsiino-nvidia]

(Based off #419 , yet to be merged)

• This updates the HF dataset upload flow to account for changes such as updated naming (e.g. we prefix with Nemotron-RL and not Nemo-Gym now), and various other things.
• Contributing files to existing repos as a PR is also supported now.
• Docs have been updated.
• Following datasets were contributed (as PRs) via:

ng_upload_dataset_to_hf \
    +input_jsonl_fpath=data/workplace_assistant/validation.jsonl \
    +resource_config_path=resources_servers/workplace_assistant/configs/workplace_assistant.yaml \
    +create_pr=true \
    +commit_message="Upload validation" \
    +split=validation

ng_upload_dataset_to_hf \
    +input_jsonl_fpath=data/structured_outputs/structured_outputs_251027_nano_v3_sdg_json_val.jsonl \
    +resource_config_path=resources_servers/structured_outputs/configs/structured_outputs_json.yaml \
    +create_pr=true \
    +commit_message="Upload validation" \
    +split=validation

ng_upload_dataset_to_hf \
    +dataset_name=OpenMathReasoning \
    +input_jsonl_fpath=data/math_with_judge/aime24_validation.jsonl \
    +resource_config_path=resources_servers/math_with_judge/configs/math_with_judge.yaml \
    +create_pr=true \
    +commit_message="Upload validation" \
    +split=validation

[copy-pr-bot]

Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually.

Contributors can view more details about this message here.

[bxyu-nvidia]

changes merged in #481