feat(server/diagnostics): device debug-reporting ingest + admin timeline + retention (M9)

New diagnostic_events table + per-account users.debug_mode_enabled flag.
When an account's flag is on, its client(s) POST a batch timeseries of
connectivity / UPnP-sync / power / lifecycle events to /api/diagnostics
(no-op 204 when off, kind whitelist mirrors the CHECK constraint).

Admin surface: GET /api/admin/diagnostics (optional account/device/kind/
time-window filters, RFC3339-or-epoch-ms, export-sized paging) + a
/diagnostics/devices overview + PUT /api/admin/users/{id}/debug-mode to
flip an account remotely while a bug is live. debug_mode_enabled is now
exposed on /api/me (client gate) and the admin user views.

Retention: a 30-day gc-worker sweep (GcPruneDiagnostics), keyed on the
server clock so a skewed device clock can't keep rows alive.

Refs Scribe M9 (#119), tasks #1172 #1173.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01K55iTxn95BtshocgdE1shW

internal/db/queries/diagnostics.sql

@@ -0,0 +1,66 @@
+-- name: InsertDiagnosticEvent :one
+-- Records one client-reported diagnostic event. The writer validates the
+-- `kind` category against its whitelist before this runs; the CHECK
+-- constraint is the belt-and-braces guard. payload carries the finer
+-- event discriminator + all event-specific fields. Returns received_at
+-- so the caller can confirm ingest.
+INSERT INTO diagnostic_events (
+    user_id, client_id, app_version, os_version, kind, payload, occurred_at
+)
+VALUES ($1, $2, $3, $4, $5, $6, $7)
+RETURNING id, received_at;
+
+-- name: ListAdminDiagnostics :many
+-- Admin timeline query. All filters are optional (NULL = no filter):
+-- account (user_id), device (client_id), category (kind), and the
+-- occurred_at window (from_ts/to_ts). Newest-first; the SPA reverses for
+-- a chronological timeline / export. Joined with users for the username
+-- so the view renders without a second round-trip.
+SELECT
+    de.id          AS id,
+    de.user_id     AS user_id,
+    u.username     AS username,
+    de.client_id   AS client_id,
+    de.app_version AS app_version,
+    de.os_version  AS os_version,
+    de.kind        AS kind,
+    de.payload     AS payload,
+    de.occurred_at AS occurred_at,
+    de.received_at AS received_at
+FROM diagnostic_events de
+JOIN users u ON u.id = de.user_id
+WHERE (sqlc.narg(user_id)::uuid   IS NULL OR de.user_id   = sqlc.narg(user_id)::uuid)
+  AND (sqlc.narg(client_id)::text IS NULL OR de.client_id = sqlc.narg(client_id)::text)
+  AND (sqlc.narg(kind)::text      IS NULL OR de.kind      = sqlc.narg(kind)::text)
+  AND (sqlc.narg(from_ts)::timestamptz IS NULL
+       OR de.occurred_at >= sqlc.narg(from_ts)::timestamptz)
+  AND (sqlc.narg(to_ts)::timestamptz IS NULL
+       OR de.occurred_at <= sqlc.narg(to_ts)::timestamptz)
+ORDER BY de.occurred_at DESC
+LIMIT sqlc.arg(lim)::int OFFSET sqlc.arg(off)::int;
+
+-- name: ListDiagnosticDevices :many
+-- Distinct devices that have reported, for the admin device filter +
+-- "who is currently reporting" overview. Optionally scoped to one account.
+SELECT
+    de.client_id            AS client_id,
+    de.user_id              AS user_id,
+    u.username              AS username,
+    max(de.app_version)::text        AS app_version,
+    max(de.os_version)::text         AS os_version,
+    max(de.occurred_at)::timestamptz AS last_seen,
+    count(*)                         AS event_count
+FROM diagnostic_events de
+JOIN users u ON u.id = de.user_id
+WHERE (sqlc.narg(user_id)::uuid IS NULL OR de.user_id = sqlc.narg(user_id)::uuid)
+GROUP BY de.client_id, de.user_id, u.username
+ORDER BY last_seen DESC;
+
+-- name: GcPruneDiagnostics :execrows
+-- Retention sweep for the gc worker. Deletes diagnostic_events older
+-- than 30 days by server clock (received_at) so a skewed client clock
+-- can't keep rows alive. Matches the other gc lifecycle sweeps'
+-- hardcoded-threshold style; diagnostics are debug telemetry, not user
+-- data, and 30 days is ample to investigate a live incident after the
+-- fact. Returns the affected row count for the sweep log line.
+DELETE FROM diagnostic_events WHERE received_at < now() - INTERVAL '30 days';