Plasma as a work queue

[Eric Talbot]

We're trying to set up a system that adds objects to plasma from 
scala/JVM, and then a separate Python process consumes the objects.

The Python piece looks something like this:
client = plasma.connect(...)client.subscribe()

while True:

object_id, data_size, meta_size = client.get_next_notification()
        if data_size >= 0:
            data, = client.get_buffers([object_id])

            func(pyarrow.BufferReader(data).read())
            client.delete([object_id])