c/archival_stm: translate sync error to not_leader error code

When `persisted_stm::sync()` method fails it is indicating that the
current node is not longer a leader. The `sync()` executed before
`replicate` call in archival stm `command_batch_builder` prevents
replicate from being called. The end result for such an error is
deterministic and we can translate the sync error to `not_leader` error
code.

Fixes: redpanda-data#14898

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

src/v/cluster/archival_metadata_stm.cc

@@ -376,7 +376,7 @@ ss::future<std::error_code> command_batch_builder::replicate() {
         auto timeout = now < _deadline ? _deadline - now : 0ms;
         return _stm.get().sync(timeout).then([this](bool success) {
             if (!success) {
-                return ss::make_ready_future<std::error_code>(errc::timeout);
+                return ss::make_ready_future<std::error_code>(errc::not_leader);
             }
             auto batch = std::move(_builder).build();
             return _stm.get().do_replicate_commands(std::move(batch), _as);