Optionally include the pipeline script in the hub when pushing your distiset

[plaguss]

Description

This PR add the option of pushing the script of the pipeline being run to the hugging face hub (by default it will be set to False, to avoid potential errors):

with Pipeline() as pipe:
    ...
distiset = pipeline.run(use_cache=False)
distiset.push_to_hub("plaguss/pipe_nothing_test", include_script=True)

This simplifies sharing the code that created the pipeline, as well as custom steps.

Example script.

If the script was uploaded to the hub, an entry will be written in the README.md of the repo to show it:

The cli has also been updated to allow running remote (or local) scripts, as we do with pipelines defined in their pipeline.yaml config file:

• Pointing to the same file we uploaded:

distilabel pipeline run --trust-code "https://huggingface.co/datasets/plaguss/pipe_nothing_test/raw/main/pipe_nothing.py"

• Pointing to the local file:

distilabel pipeline run --trust-code "path/to/pipe_nothing.py"

[codspeed-hq]

CodSpeed Performance Report

Merging #762 will not alter performance

_{Comparing push-pipeline-py (cbcf0bf) with develop (647d040)}

Summary

✅ 1 untouched benchmarks

[gabrielmbmb]

LGTM!