After a named halt duty attaches to one finding class, which pre-deployment evaluation still runs on systems where that class might fire?

For one named domestic actor in 2005–2007, did housing-leverage warnings fail because the seat could not bind, or because it could bind and bore no cost for not binding?

Who may set, and who may refuse, the threshold that makes a finding a trigger — and who may contest it without turning the route into denial-of-service?

After a published inadequacy assessment, does any actor have to impose common rules, or record a reason for not imposing them, inside a deadline?