BalancerCommandsScheduler is currently storing in config.balancerCommandsSchedulerOngoingOperations the information on commands to be re-issued after a step-down event (similarly to what the MigrationManager did with config.migrations, but using a different schema).

This may lead to some issues, such as:

• outstanding moveChunk commands stored in config.migrations that won't be recovered after a server upgrade
• outstanding moveChunk commands stored in config.balancerCommandsSchedulerOngoingOperations that won't be recovered after a server downgrade
• stale moveChunk being wrongly re-issued after a downgrade+upgrade cycle.

The simplest way to solve this problem is to avoid interactions with the FCV routine - and to just keep using `config.migrations` for the task.