The PoV benchmarking aims to estimate the worst-case scenario, which is unlikely to occur on average.
Too large over-estimations hurt block utilization. There is currently no way to retroactively measure the actual used proof size.
Having such a retroactive measurement in place would enable a multi iteration approach; alternating between dispatch and assessment phase to reach maximum block utilization.

It would therefore be nice to have a way to measure the weight of a batch of dispatchables after their execution. For example in:

• Scheduler Task dispatch
• XCM dispatch
• Transaction dispatch
• MessageQueue processing

Basically in every situation where we need to fit a batch of Dispatchables into a weight limit, and assume that the pre- and post-dispatch PoV weight differ greatly.