Examine produced optimised weights under common `reweight` function

We'll want to run some exploratory analysis on what happens when we move the UK and US calibration to the same reweighting routine- things like:

* What happens to the sample size in each country?
* Do both work OK?
* How does loss change in each country?