Ability to execute migration only once on the single node of a cluster

Now all migrations will execute on each node of the cluster.

I want to insert some initial data to some sharded space in a cluster (constants, configs etc).
For example consider the following migration code:

return {
    up = function()

        crud.insert('dictionary', {1, box.NULL, 'en_ru', 'apple', 'яблоко'})
        crud.insert('dictionary', {2, box.NULL, 'en_ru', 'tomato', 'томат'})

        return true
    end,

    run_on_single_node = true
}

It seems that if such migration completes without errors it must be marked as 'EXECUTED' on all nodes.

No files, tests, or entry points are named. Start by tracing how migrations execute on cluster nodes and how completed migrations are recorded, then verify the behavior with a multi-node migration that inserts initial sharded data. Done means the migration runs on one node while its completion is recorded consistently on all nodes.