Evaluate scoped-hash gate vs sibling-clone for solo-pluto push isolation

Fleet solo-agent brainstorm bs-mr50y0va70m (2026-07-03): mars proved the lighter fix for solo push-pollution is scoping the Class-A gate HASH to solo's own paths, NOT a sibling clone. Pluto's .gitpush-pre.sh:46 hashes WHOLE-TREE (git diff HEAD -- .); coder/db dirty WIP poisons solo's verdict-hash -> the PLUTO-444 stash-dance. Pluto currently solves this with an own-clone (pluto-solo/, built+proven+live). This WI = evaluate switching to scoped-hash (retired PLUTO-475 manifest-gate idea) IF the clone's per-WI git pull --rebase + .env copy + node_modules sync overhead proves painful. Clone stays as-is until then (don't rip working infra). Scoped-hash is the agreed FLEET DEFAULT for NEW solos. someday-tier.