Description of problem:
When we run tests with the Reboot step in NMCI it sometimes needs a second to be ready to run commands like nmcli device but in other cases, it needs 10+ seconds to be ready. We do run assert nmci.nmutil.start_NM_service(timeout=timeout), "NM start failed" to have this ready but this is not helping (and we had to bump timeout from 5 to 10s even for that)

I have a test case showing this when running in cycle stored in vb/failing_reboot (instruction how to run that are in commit message)

Version-Release number of selected component (if applicable):
NetworkManager-1.40.9-31044.copr.63a8cec1b1.el9.x86_64

How reproducible:
randomly but usually we can see it in 200 runs

Steps to Reproduce:
1. get NMCI
2. test=test
3. a=0; while ./test_run.sh $test; do :; sleep 2;((a++)); echo "ATTEMPT $a"; if [ $a -eq 200 ]; then break; fi ; done; echo "ATTEMPT $a"

Actual results:
stability issues

Expected results:
rock solid and quick service restarts

Additional info:
we probably need to report NM service as down and up after some checks are performed and we are sure other commands can run successfully

Acceptance Criteria:
Given a VM with NetworkManager running 

When the integration tests with the reboot steps are run

Then the service should quickly restarts