ci: fix stress counter in run'n'monitor
The summary about the stress test often finish with some results like:
succ: 0; fail: 0; total: 2 of 2
So, it missed to correctly counting the retries due to the --stress
test.
Checking that, what happens is the accumulation happens after the enable_job(..."retry"...)
and the job.status
stored is the new one, but not the completed one. Just moving this before the retry, I think solves the output problem.