Create a reusable file loader component

[RobbeSneyders]

It would be useful to have a reusable file loader component which can load data from files in a directory.

It should translate the following directory structure:

|- data_directory
   |- filename_1
   |- filename_2
   |- filename_3

To the following Dask dataframe:

id	file_content
filename_1	file 1 content
filename_2	file 2 content
filename_3	file 3 content

A minimal component specification would look like:

name: Load from files
description: Component that loads a dataset from files
image: ghcr.io/ml6team/load_from_files:dev

produces:
  file:
    fields:
      content:
        type: binary

args:
  path:
    description: Local or remote path to the directory containing the files
    type: str

Some notes:

• We can make the component "generic" as a next step (meaning it can handle dynamic component specifications)
• Remote paths should work out of the box if the component installs fondant[aws,azure,gcp].
• We can add additional functionality later (eg. handling subdirectories, taking additional arguments, ...

[satishjasthi]

@RobbeSneyders I'll take up this issue

[satishjasthi]

@RobbeSneyders, For downloading files from a remote path, what options are we considering now?
Should it support downloading files from Cloud computing platforms like GCP, AWS and Azure? If so should i add component which handles auth part to fetch directory data and then parse it.
Or should the current version of code just downloads remote dir in the form of zip or tar and then unpacks it and then parses it

[satishjasthi]

@RobbeSneyders I have raised PR for this issue