Fix minor memory leak in clusterLoadConfig (#741)

We forgot to call sdsfreesplitres in the error path during a
nodes.conf corruption check, this function exits on the error
paths so this is just a cleanup.

Signed-off-by: bentotten <59932872+bentotten@users.noreply.github.com>

src/cluster_legacy.c

@@ -578,6 +578,7 @@ int clusterLoadConfig(char *filename) {
                        memcmp(primary->shard_id, n->shard_id, CLUSTER_NAMELEN) != 0) {
                 /* If the primary has been added to a shard, make sure this
                  * node has the same persisted shard id as the primary. */
+                sdsfreesplitres(argv, argc);
                 goto fmterr;
             }
             n->replicaof = primary;