Hosts/nodes console: offline hosts read Ready

Found while diagnosing aws1 (Task 5573): its daemon was wedged on a hung EFS mount for 90 minutes. The console made that hard to see. Requirements: 1. The host page (`/nodes/hosts/<host>`) shows "Ready" while its only node and its host node have been unseen for 30+ minutes. Host status must turn offline/stale when its nodes do, using the presence verdict the node page already uses. 2. The `/hosts` route renders "Something went wrong loading this page". Fix the crash, or redirect it to `/nodes?tab=hosts`. 3. A stale "Add node failed" banner sits on the host page with no date or reason. Show when it failed and why, and let the operator dismiss it once a node is healthy. 4. The node Health tab shows only that every reporting clock is STALE. When the daemon logs `did not answer a mount check` / `periodic step … wedged`, surface that as a Health finding ("NodeLink store not answering (NFS)") so a hung mount is named rather than inferred. 5. The node page gives no way to reach the host (no SSM/terminal link) and no cloud instance id. Show the hosting instance id, region and account for a Jaah-hosted host. Operators currently need the AWS CLI to find the machine. Out of scope: the lane/placement and networkd fixes (Task 5573 PR). Done means: each item verified on the live console with a screenshot in the report.

All 5 items merged (#5247 + #5254 /hosts org-scope redirect fix), full gate green, Jaah Deploy ok, verified live on dev except item 4 (no live hang; verified locally). 5 UI follow-ups named in report (task creation refused).