Add tenant-aware capacity planning for the shared database

[JSONbored]

Problem: The shared production database already hit its account-level storage cap once from unbounded row growth at today's single-operator scale — a real, already-observed ceiling that a multi-tenant model would hit sooner without deliberate planning.

Area: ORB / Cloud

Proposal: Add per-tenant storage accounting and proactive alerting before any multi-tenant rollout, building on the retention fix already shipped for the incident that occurred.

Deliverables:

• Per-tenant storage visibility and an alert before the shared account-level cap is approached again.

[linear-code]

JSO-3550

[JSONbored]

Re-scoping — the account-wide alerting half of this already shipped: src/selfhost/d1-size-probe.ts + prometheus/rules/alerts.yml (LoopoverD1DatabaseSizeWarning/Critical at 70%/90% of the 10GB cap) already exist, citing #3810's incident. What's still genuinely missing: a per-installation breakdown of that storage — today's alerting is account/table-level only, no way to attribute usage to a specific tenant. Narrowing this issue to that gap specifically.