[Feature Request][prism]: Add support for the multi-chunk iterable protocol

[lostluck]

What would you like to happen?

Prism's goal is to simplify testing all the bits of beam, and it has it's first real use for this!

#23043 revealed that the Spark runner uses the "multi-chunk" iterable protocol, which isn't commonly used outside of the state backed iterable path.

https://github.com/apache/beam/blob/master/model/pipeline/src/main/proto/org/apache/beam/model/pipeline/v1/beam_runner_api.proto#L859

This path wasn't properly tested for the singleIterate path in the Go SDK, revealing that the iterator being left undrained put the reader in a bad position for the next element, leading to a bug.

Prism should implement the multi-chunk iterable behavior, and have a means for configuring the iterable behavior from a passed in pipeline option.

The validation for this will come from tests in the Go SDK's exec package and validating that those paths are covered.

Issue Priority

Priority: 2 (default / most feature requests should be filed as P2)

Issue Components

• Component: Python SDK
• Component: Java SDK
• Component: Go SDK
• Component: Typescript SDK
• Component: IO connector
• Component: Beam examples
• Component: Beam playground
• Component: Beam katas
• Component: Website
• Component: Spark Runner
• Component: Flink Runner
• Component: Samza Runner
• Component: Twister2 Runner
• Component: Hazelcast Jet Runner
• Component: Google Cloud Dataflow Runner