There are some specific MachineConfigs which are rebootless and my understanding is that this is hardcoded.

Many configurations applied from MachineConfigs do not require a massive reboot of all the nodes in the pool. Therefore, my suggestion is that there is a rebootless field (or any other descriptive name) on MachineConfig objects whereby users can decide and configure whether the MachineConfig shall reboot the nodes of the pool or not.

The partner which is requesting this has specific reasons and they will be described in a private comment. However; I also think that avoiding unnecessary reboots would be optimal for any user. Anyway, if I am missing any detail or you think this reasoning is wrong, please feel free to let me know.

This feature would imply that users must understand whether a node reboot is needed for the configuration that they would need to apply. Therefore, the default behavior could be that the reboots happen unless it is configured the opposite.