Support for Multiple URLs by APIDataSet

[Minyus]

Description

The provided APIDataSet accepts only a single URL.
I'd like to request support for multiple, potentially a lot of, URLs similar to PartitionedDataSet interface, without being interrupted even if errors occur for some of the URLs.

Context

It is useful to get multiple contents (e.g. json, image, video, model, zip files) through API.
Kedro provides PartitionedDataSet to support multiple datasets in a dict, but cannot be used for this purpose because:

• It is possible that some of the URLs return errors while the others work fine. An option to skip errors and get contents as many as possible will make sense.

• It is inefficient to configure a TCP connection for every requests.request call. Instead, requests.Session allows to reuse the same TCP connection to handle multiple requests in less time.

Possible Implementation

1. Accept a dict of URLs similar to PartitionedDataSet.
2. Use requests.Session in __init__ method.
3. Use methods of the requests.Session object to send requests.
4. Return dict of contents using the same keys as the dict of URLs. For unsuccessful requests, return Exception/response objects without raising.

[Minyus]

I implemented this feature and published at: https://github.com/Minyus/pipelinex/blob/master/src/pipelinex/extras/datasets/requests/api_dataset.py#L168