[CB] Refactor logic for inputs and outputs outside of the main API

[remi-or]

This PR refactors all of the logic linked to the inputs and outputs of continuous batching into a new class ContinuousBatchingIOs itself in a new file. This will be quite useful when we introduce async scheduling, because there will be two sets of host-side IOs and one set of device-side IOs.
To avoid having this major refactoring AND the async logic in the same PR, I think it would be good if we can first merge this part, which does not change much, and then add the async logic.
This PR also makes the test file ty compatible and adds a bit more documentation.

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

[ArthurZucker]

this makes the code a bit more stateful, but it might be better. I am mostly thinking of multi-gpu runs, this should make stuff simpler.