feat: add sync checkpoint restore for last successful sync

Capture rolling online DB snapshots after push/drain/reconcile and expose SyncPage 数据恢复 plus checkpoint/restore APIs; mark Z12h and restore done in coop docs.

Co-authored-by: Cursor <cursoragent@cursor.com>

platform/internal/handler/agent_sync.go

@@ -90,6 +90,9 @@ func agentSyncPushHandler(svcCtx *svc.ServiceContext) http.HandlerFunc {
 		if res != nil && res.Skipped {
 			outcome = "skipped"
 		}
+		if res != nil && res.Applied {
+			dbsync.ScheduleCheckpoint(svcCtx.DBSync.Store(), ch, "push")
+		}
 		logSyncReq(r, "push", channelID, item.Table, item.RowPK, outcome, "", dur, reqID)
 		auditSync(svcCtx, r, "dbsync.push", map[string]any{
 			"channel_id": channelID, "table": item.Table, "row_pk": item.RowPK,
@@ -153,6 +156,15 @@ func agentSyncPushBatchHandler(svcCtx *svc.ServiceContext) http.HandlerFunc {
 			httpx.OkJson(w, payload)
 			return
 		}
+		applied := 0
+		for _, r0 := range results {
+			if r0.Applied {
+				applied++
+			}
+		}
+		if applied > 0 {
+			dbsync.ScheduleCheckpoint(svcCtx.DBSync.Store(), ch, "push_batch")
+		}
 		logSyncReq(r, "push_batch", channelID, "", "", "ok", "", dur, reqID)
 		auditSync(svcCtx, r, "dbsync.push_batch", map[string]any{
 			"channel_id": channelID, "n": len(body.Items), "ms": dur.Milliseconds(), "req_id": reqID,