[feat] Add a batch validation mode

[wylansford]

Description
Some users are queuing results and wish to be able to batch a validation request to the Guard (with async support for this call, reliant on #832)

Why is this needed
In prod systems, validation latency may be too high to run it on every request. Batching makes more sense, and should be supported

Implementation details

End result

[CalebCourier]

@wylansford would this necessitate batching inference calls as well? I'm thinking that's likely where the majority of any latency derives from within the validation loop, but that's pure speculation right now.