nvidia-migmanager reboot-if-required is called (and fails) in nvidia variants

**Platform I'm building on:**
nvidia variants in AWS

**What I expected to happen:**
no errors

**What actually happened:**
journal logs
```
Aug 19 19:36:26 ip-172-31-2-139.us-west-2.compute.internal nvidia-migmanager[2493]: 19:36:26 [INFO] nvidia-migmanager started
Aug 19 19:36:26 ip-172-31-2-139.us-west-2.compute.internal nvidia-migmanager[2493]: Failed to read settings from config at /etc/nvidia-migmanager/nvidia-migmanager.toml: No such file or directory (os error 2)
```

**How to reproduce the problem:**
Boot an nvidia variant AMI

This is due to the drop-in being added for nvidia variants. We can put it in only in case of nvidia k8s variant.

Start by locating the nvidia-migmanager drop-in and the variant configuration that enables it, then compare the regular nvidia variants with the nvidia Kubernetes variant. Boot an affected AMI and verify that nvidia-migmanager no longer tries to read the missing configuration or reports the failure in the journal.

Written by the indexing model from the issue text.