Should we publish performance benchmarks (latency/accuracy) for each guard so users know which one to pick?

[vaibhav-research]

to make it production ready, I think we should explore publishing key metrics to help users decide which one is good to pick.

[dronefreak]

Report latency / memory footprints / dependancy tree for each defence
Segregate based on user requirements (fast and accurate, slow and very accurate etc.)