Batch Job Support

We will be enabling batch job support by creating an adapter which will work for Functions of all languages without needing to alter the functions themselves. The user will indicate the Function should be a batch processor, and it will be deployed as a JobSink. Functions will then deploy in tandem a job adapter which reads the job to be processed and invokes the Function with this job as the content of an event. In this way we can adjust how a function is invoked (standard HTTP requests, cloudevent, or a batch job cloudevent) without needing to alter the Function implementation itself. Note this may only support CloudEvent functions.

Start by surveying the existing function deployment and invocation paths, then trace how JobSink resources and CloudEvents are currently represented. Define the adapter and deployment boundaries described here, and verify that batch processing can be selected without changing function implementations and that supported invocation modes remain intact.

Written by the indexing model from the issue text.