I suggest you click Report to Moderator and ask to have this Thread merged with your earlier one because the context of the problem is set out there and it is very important in this case.
In general my master to many slaves code assumes that all of the slaves are always listening. If you put a slave to sleep it will likely miss a call from the master. I think it is possible to have the nRF24 wake the slave when a message arrives but I have never experimented with that. And for that to happen the nRF24 could not be asleep.
The alternative is to make the master the listener and have the slaves wake up and call it at intervals. But that will need careful coding in the slave to deal with data collisions when two slaves transmit at the same time. The auto-retry feature is no help in this case, it will probably make things worse. So that the master knows where a message comes from each slave should include an identifier as part of the message.
You could also manage things so that every slave wakes for long enough to hear a message if the master sends one. If, for example, a slave is awake for 1 second every 10 seconds that would still be a 90% energy saving.
...R