Losing the runner slows the gate and never blocks a release of a fix #1196

Part of #1148 (failure domain: the runner). When checks cannot run (the cell is gone, the machine is out of memory, the queue does not move), a pull that fixes something must still be landable by a person and releasable.

Done looks like:

• [ ] A check that cannot start in a bound time says so on the pull as "the runner is not answering", never as red, and vitals names it.
• [ ] A person with admin can land a pull whose check is unavailable, with the reason recorded on the pull; a red check is still refused.
• [ ] A test takes the runner away and proves both.

Parked by the crew: the check failed and the decider saw no point in another try (pull #1201 closed). Planner: split this smaller, or rewrite what done looks like; a person can also say what to try.

Parked by the crew: the check stayed red through 2 automatic tries (pull #1522 closed).

Parked twice since a person last spoke here, so it is the planner's now. Every reason:

1. the check failed and the decider saw no point in another try (pull #1201 closed)
2. the check stayed red through 2 automatic tries (pull #1522 closed)

The planner splits it, replaces it with a fresh issue, or defers it; a person can also say what to try. Nothing starts here until then, and the board lists it.

Deferred behind the Later gate (#595): a runner-loss fallback waits behind /flow (#1554), which is what will show the owner a stalled check queue first.