c/md_dissemination: fixed retrying leadership dissemination

After change from `std::vector` to `ss::chunked_fifo` to keep the
updates in metadata dissemination service `update_metadata` we lost
ability to retry metadata dissemination requests as pending updates were
moved to lambda capture list. This way the next retry did not have any
updates to deliver.

Signed-off-by: Michal Maslanka <michal@redpanda.com>

src/v/cluster/metadata_dissemination_service.cc

@@ -410,13 +410,19 @@ ss::future<> metadata_dissemination_service::update_leaders_with_health_report(
 
 ss::future<> metadata_dissemination_service::dispatch_one_update(
   model::node_id target_id, update_retry_meta& meta) {
+    // copy updates to make retries possible
+    ss::chunked_fifo<ntp_leader_revision> updates;
+    updates.reserve(meta.updates.size());
+    std::copy(
+      meta.updates.begin(), meta.updates.end(), std::back_inserter(updates));
+
     return _clients.local()
       .with_node_client<metadata_dissemination_rpc_client_protocol>(
         _self.id(),
         ss::this_shard_id(),
         target_id,
         _dissemination_interval,
-        [this, updates = std::move(meta.updates), target_id](
+        [this, updates = std::move(updates), target_id](
           metadata_dissemination_rpc_client_protocol proto) mutable {
             vlog(
               clusterlog.trace,