Dashboard benchmark view: gate precision vs peer median for federated fleet intelligence

[JSONbored]

Context

#1970 (opt-in federated fleet intelligence) calls for "a 'your gate precision vs peer median' benchmark in the dashboard when federation is enabled." Once export (#6478), the collector client (#6479), and import+validation (#6480) are landed, an opted-in instance has real local calibration data plus verified peer bundle data to compare against — this issue is the dashboard surface that renders that comparison. Blocked by #6478 and #6480 (needs both a working export path and a working, trust-gated import path to have real data to render against — building the view against fabricated data first would produce a UI nobody can verify against the real pipeline).

Requirements

• Only renders when federation is enabled for the instance (.loopover.yml toggle from Implement signed, anonymized signature-bundle export for federated fleet intelligence #6478/Opt-in collector client for federated fleet intelligence (push/pull transport) #6479/Implement signature-bundle import + trust-gating validation for federated fleet intelligence #6480) — an instance that hasn't opted in sees no new UI, not an empty/disabled version of it.
• Shows this instance's own gate precision alongside the peer median computed from imported, trust-gated bundles only (never raw/unvalidated peer data).
• Follows this repo's general preference for reusing existing dashboard chrome over adding new pages — place this as a new panel/section within the existing maintainer dashboard (see apps/loopover-ui/src/components/site/app-panels/maintainer-panel.tsx for the existing panel convention) rather than a standalone route, unless the data shape genuinely doesn't fit there.
• Bearer-gated, maintainer-only visibility (this is an operator's own private dashboard, not a public surface) — matches the access model already used by other maintainer-only panels in this codebase.
• No wallet/hotkey/reward/trust-score wording — matches this repo's existing public/private-surface wording constraints even though this view itself is private, for consistency with the rest of the dashboard's vocabulary.

Deliverables

• A new dashboard panel/section rendering local-precision-vs-peer-median, gated on the federation opt-in toggle and bearer auth.
• Loading/empty states for an opted-in instance with zero peer data yet (not an error state).
• Tests covering the gated-visibility logic (opted-out renders nothing, opted-in-with-no-peers renders the empty state, opted-in-with-peers renders the comparison).

Test Coverage Requirements

This is UI work under apps/loopover-ui/**, outside coverage.include per this repo's convention — Codecov's patch gate does not apply to the UI component itself. If this issue adds any new src/** aggregation/query logic to compute the comparison server-side, that code inherits the normal 99%+ patch requirement.

Expected Outcome

An operator who has opted into federation sees a real, live comparison of their own gate precision against the peer median, sourced only from trust-gated, verified peer data.

Links & Resources

• Opt-in federated fleet intelligence across self-hosted instances #1970 (parent — federated fleet intelligence)
• Implement signed, anonymized signature-bundle export for federated fleet intelligence #6478 (export — this instance's own data), Implement signature-bundle import + trust-gating validation for federated fleet intelligence #6480 (import+validation — the verified peer data this view compares against)
• apps/loopover-ui/src/components/site/app-panels/maintainer-panel.tsx (existing panel convention)

[linear-code]

JSO-4447