For our campaigns, we have used CBO in an attempt to loosely A/B test audiences and creative sets for a few of our shorter running campaigns. Specifically, setting the chunk of dollars at the campaign level and then putting minimum spends in each of ad sets to ensure that a small percentage (like 10-20%) will be spent on each set and then allow Facebook's algorithm to spend the rest accordingly.
We started this when a few different clients didnt "care" what creative or audience was specficially supported, they just wanted the spend to do well.
For the second part of your question, we have used a very similar structure on CBO at a high frequency. We have also swtiched our "broad" with a different version of a lookalike (1st or 3rd party data) to understand which group would drive the stronger results. As long as you set a minimum spend to get everything a fair shake, I would recommend using this structure.