Our python functional testing framework is pretty limited for what kinds of p2p behaviors we can test. Basically, we can currently only make manual connections between bitcoind nodes (using the addnode rpc), which are treated differently in our code than outbound peers selected using addrman.

While we do have some unit-testing coverage of some of the components (like addrman, and parts of net_processing), I don't believe we currently are able to test the overall logic of how bitcoind uses those components (I recall this coming up when working on #11560, as a specific example).

Anyway I am just mentioning this here as a potential project idea, as this is a material gap in our testing that I think would be valuable to work towards improving, and I wasn't sure how well known this is.