fix(sync): notify declarative:reconfigure events for full sync (#13860)

With full sync (`wipe = true` in `do_sync()`), if we dont notify other workers with `declarative:reconfigure`, the registered callbacks in other workers will not be called, like updating `kong.default_workerspace`.

Note that when an empty-configured CP and DP start with incremental sync enabled, the DP will rebuild routers, plugins, and the balancer twice. This occurs because the DP currently calls do_sync() twice during a single sync operation. With incremental sync disabled, it rebuilds these components only once.

https://konghq.atlassian.net/browse/KAG-5812
(cherry picked from commit 945d333)

kong/clustering/services/sync/rpc.lua

@@ -14,6 +14,7 @@ local declarative = require("kong.db.declarative")
 local constants = require("kong.constants")
 local concurrency = require("kong.concurrency")
 local isempty = require("table.isempty")
+local events = require("kong.runloop.events")
 
 
 local insert_entity_for_txn = declarative.insert_entity_for_txn
@@ -341,6 +342,16 @@ local function do_sync()
     kong.core_cache:purge()
     kong.cache:purge()
 
+    -- Trigger other workers' callbacks like reconfigure_handler.
+    --
+    -- Full sync could rebuild route, plugins and balancer route, so their
+    -- hashes are nil.
+    local reconfigure_data = { kong.default_workspace, nil, nil, nil, }
+    local ok, err = events.declarative_reconfigure_notify(reconfigure_data)
+    if not ok then
+      return nil, err
+    end
+
   else
     for _, event in ipairs(crud_events) do
       -- delta_type, crud_event_type, delta.entity, old_entity