[FEATURE] Create a direct way of passing a dataset as part of a pipeline

[plaguss]

Is your feature request related to a problem? Please describe.
There are users that want to process a dataset on its own and just use it in a pipeline, without writing it to disk first, or uploading it to the Hugging Face Hub and download it afterwards. We have to offer this possibility.

Describe the solution you'd like
A simple way of passing a datasets.Dataset or pandas.DataFrame for example already loaded to a Pipeline.

Describe alternatives you've considered
Write it first to disk or upload it to the HF Hub.

Additional context
Add any other context or screenshots about the feature request here.

[dvsrepo]

For ref, this is how I work to iterate a sample of my data:

https://huggingface.co/blog/dvilasuero/synthetic-data-with-llama3-distilabel#load-dataset

# get great prompts annotated by at least 2 contributors
dataset = load_dataset(
  "DIBT/10k_prompts_ranked",
  split="train"
).filter(
  lambda r: r['avg_rating']>=4 and r['num_responses']>=2
)

dataset = dataset.to_list()
load_dataset = LoadDataFromDicts(
  name="load_dataset",
  data=dataset[0:500], # during development I used [0:1] to iterate quickly
  output_mappings={"prompt": "instruction"}
)