Description of problem:
I was able to stop nvmeof gateways one by one that were under load without being 'challenged' by cephadm. With the maintenance command, we run "ok to stop" logic, so I think that would also be good to implement in the orch daemon stop - at least for the gateways.

Note that any "ok to stop" logic would also need to account for multiple gateway groups.

Version-Release number of selected component (if applicable):
Ceph 7.1 (18.2.1-149.el9cp)

How reproducible:
every time

Steps to Reproduce:
1. Create a cluster with multiple gateways in a group
2. Use ceph orch daemon stop <name> to stop the nvmeof gateways one by one
3.

Actual results:
All gateways are stopped, without the requirement for a --force parameter 

Expected results:
stopping all gateways in a group should at least need a --force parameter for the last gw in a group.

Additional info: