Draft: Set enable stale production flag on one node

instead of correcting timestamp for faketime

We could use notification system to wait until node starts, but changing faketime during runtime doesn't solve problem (block producing code spends most of time in sleep system call and kernel is not aware of faketime). And as I checked code - this flag is used only during node startup before first block in live sync appears. Parameter witness_partitipation_rate is different, its used at every block production opportunity and we still need faketime to keep default value of this parameter.