Host stuck joining shows blank status

Found while diagnosing aws-gmind-2 in prod (GMind org): the host installed fine (jaahd 1.2.4936, Healthy, reporting since 20:20 UTC on 2026-10-27), and its node aws-gmind-2 was placed at 20:22. The node never reported in ("1 issue: silent", Slots 5, Last Seen None). The host's Status reads only "No progress in 14 min." Nothing in the UI says why: no converge error, no step, and the host's Config snapshot clock is STALE from 20:21. An operator without a shell on the box can't go further. Requirements: 1. While an order is in `joining`, the host page names the step the host's converge is on for the node, and the host's last converge error if there is one, in place of the bare "No progress in N min.". 2. When the host's config snapshot is stale past the stall threshold, the host page says so in one line ("Host stopped reporting its config N min ago") with the jaahd hint, as the connected-not-polling case already does. 3. Root-cause why aws-gmind-2 (EC2 m8i.large, Ubuntu 26.04, 2 vCPU / 7.6 GB, self-hosted via SSH paste) never brought its node up, and fix it if it's a product bug. Out of scope: the self-hosted install-command / EC2 user-data shebang issue (fixed by Task 10757). Done means: on a host whose node is stuck joining, the Status tab shows a concrete reason or step instead of only "No progress in N min."; hostFacts tests cover it.

Merged #7168 (5c0e896). A host stuck joining now shows its converge step and the host's last error, which first-boot holding used to hide. After 16 min without a poll it adds "stopped reporting" and the jaahd hint. Quick gate and changed-area tests green; both review lanes passed. Couldn't SSH into aws-gmind-2: the key was refused. Memory: none applied.