[Story]: Privacy aggregation-floor policy layer

## Summary
Enforce a minimum-cohort aggregation floor uniformly across every connector, using Copilot's 5-seat rule as the reference minimum, so that small teams cannot be de-anonymized by cross-referencing adoption/acceptance data. This must land alongside the first connectors, not retrofitted afterward.

## Intent / Source of truth
2026-08-13 backlog consolidation: "we are an AI-governance vendor and must model the discipline we sell." Copilot's own 5-seat-per-day minimum (story 3) as the reference floor.

## Scope
- [ ] A shared, connector-agnostic aggregation-floor policy layer (minimum cohort size before any per-user-adjacent data is surfaced)
- [ ] Applied uniformly to every connector from day one — not vendor-by-vendor opt-in
- [ ] Documented policy (what the floor is, why, and what happens when a cohort is below it — e.g. suppressed panel vs. rolled-up-to-parent-team)

## Out of scope
- Per-vendor custom floors that diverge from the shared policy without a documented reason
- Retroactive redaction of already-shipped dashboards (this should land alongside, per the "not afterwards" requirement — if any connector ships before this, that's a defect to flag, not accept)

## Verification
A test cohort below the floor shows suppressed/rolled-up data, not raw small-N numbers, across at least two different connectors' dashboards.

## Risk assessment
This is a trust/positioning risk for an AI-governance vendor specifically — shipping any connector without this floor in place first would be a credibility problem, not just a technical gap. Treat "lands alongside the first connectors" as a hard sequencing requirement.

## AI Usage Declaration
Drafted with AI assistance during the 2026-08-13 cross-repo backlog consolidation. A human owns intent, verification and consequences; acceptance criteria must be verified against the live system before this is closed.

No implementation files or test entry points are named. Start by locating the first connectors and their dashboards, then trace where cohort data is surfaced. Done means a shared floor is applied before connector data ships, the below-floor behavior is documented, and a test cohort is suppressed or rolled up across at least two connectors.

Vom Indexierungsmodell aus dem Issue-Text verfasst.