1. They are most probably for Tagging purpose, but that still needs a clarification from our product team. gmathur1133 Can you get this confirmed from engineering or product teams? I have not seen any difference in data collection or segmentation while using any of these options.
2. Person ID can be used when the trait/data source etc. belongs to a personal device example: mobile phone. Household can be used when there is a shared device example: TV, iPad etc.
Not quite sure if I understand the first question, but can provide some input on the second one. The way we've used it in one the cases was something like:
Say for example, you've an online streaming website like Netflix. There's one account and then there are multiple profiles within it. In that case, you'd most likely have an account ID associated with the account and profile IDs associated with each profiles. So there'll be two cross device data sources, on at account id level and the other at profile id. Account one here would be at Household and Profile at person level.