Add `http_dataset` Prefix and `s3` Support to `registry.json`

[ubyndr]

Description

We would like to enhance the registry.json to support additional dataset types by adding a prefix for HTTP-based datasets and including support for S3 URLs.

Proposed Changes:

1. Add a Prefix for HTTP Datasets (http_dataset)

   • Add a new prefix for HTTP datasets to the registry.json. This will allow for downloading datasets without relying the census API.
   • Example value:

     "http_dataset": "https://datasets.cellxgene.cziscience.com/xyz.h5ad"

2. Add Support for S3 Datasets

   • Include S3 dataset URLs in the registry.json for handling datasets stored in S3.
   • Example value:

     "s3": "s3://bucket_name/path/to/dataset.h5ad"

Why this is needed:

• The http_dataset prefix will allow for a more structured and modular handling of datasets that are retrieved via HTTP.
• Adding support for S3 datasets ensures that datasets hosted on AWS S3 can be seamlessly integrated into the registry, aligning with current data storage practices.

Additional Context:

• This change will ensure consistent handling across different downloaders (e.g., CxGDownloader, HTTPDownloader, S3Downloader in cas-tools) and make it easier to manage datasets in the registry.

cc @dosumis , @hkir-dev