There are several scenarios in which we'd like to be able to abort / stop all other patches from landing until a given PR has made it through. Including:

• Easily bitrotted but high-value PRs (e.g., updates to lots of key Rust components)
• Massive refactorings
• A backout of change that is tree-breaking

I'd propose something like r+ p=tree-closing. There are at least two parts to this:

1. The internal mechanics of ensuring that no other PR lands (and anything in-progress is aborted, à la force) while something with this priority is in the queue
2. Ensuring the queue display page clearly reports the reason that the queue is not processing, especially when something like a build failure happens in that PR but the queue is still blocked

cc @edunham @Manishearth @globau

(I'd especially like bikeshedding on the name of the priority, as I'm sure there's something pithy but I'm missing it)