audit: fix issues #1-#9 (mock data, notifier interface, retry/timeout, digest scheduling, real status, RPC retry, dedup, address edit UI)

- #1: remove mock alert events that poisoned production responses
- #2: add Notifier interface + Send implementations for all channels
- #3/#4: notification goroutines now carry a 30s context timeout and retry up to 3x with exponential backoff
- #5: schedule email digest in poller via DIGEST_INTERVAL_HOURS (default 24h)
- #6: SystemStatus queries real checkpoint data; returns starting/active/idle with latestBlock and lag
- #7: Ethereum RPC client retries on network errors, 429, and 5xx with 1s/2s backoff
- #8: alert_events dedup index + ON CONFLICT DO NOTHING to prevent duplicate events on poller restart
- #9: PATCH /v1/addresses/{id} for label editing; frontend address list gains inline edit and remove buttons

backend-go/internal/models/checkpoint.go

@@ -3,6 +3,7 @@ package models
 import (
 	"context"
 	"errors"
+	"time"
 
 	"github.com/jackc/pgx/v5"
 	"github.com/jackc/pgx/v5/pgxpool"
@@ -17,6 +18,20 @@ func NewCheckpointModel(pool *pgxpool.Pool) *CheckpointModel {
 	return &CheckpointModel{pool: pool}
 }
 
+// GetLatestBlock returns the highest last_checked_block and its timestamp across all addresses.
+// Returns nil, nil, nil when no checkpoints exist yet.
+func (m *CheckpointModel) GetLatestBlock(ctx context.Context) (*int, *time.Time, error) {
+	var block *int
+	var checkedAt *time.Time
+	err := m.pool.QueryRow(ctx,
+		`SELECT MAX(last_checked_block), MAX(last_checked_at) FROM address_checkpoints`,
+	).Scan(&block, &checkedAt)
+	if err != nil {
+		return nil, nil, err
+	}
+	return block, checkedAt, nil
+}
+
 // GetLastCheckedBlock returns the last checked block for an address, or -1 if never checked.
 func (m *CheckpointModel) GetLastCheckedBlock(ctx context.Context, addressID int) (int, bool, error) {
 	var block int