[FEATURE] Add Caption Queue and Caption Batching

[fearnworks]

Is your feature request related to a problem?

• Currently batch captioning is only available in the offline pipelines and not through the ui. We should be able to schedule batch captions, view the current queue, and edit the caption queue/ cancel jobs.

Describe the solution you'd like

• Add capability to batch multiple captions
• Add capability to view the queue
• Add capability to batch by image (i.e perform these captions for this image)
• Add capability to batch by perspective (i.e run this perspective on this dataset)
• Ensure UI updates as captions come in

[fearnworks]

Need to iron out some more runway for this one. Did some extensive prototyping with RabbitMQ to manage the batch queues and did not like how many abstractions were being layered on, however we still need some way of orchestrating the decentralized compute and providing incremental results to the client. Leaving issue, but blocking on upstream work for now.