Warn when a job should start but is still running

[krallin]

This will make it easier for users to debug hung jobs: rather than
requiring users to go back to when the job started to realize
(implicitly) that it's not running, this will warn at every interval
that something is wrong.

---

cc @fancyremarker @UserNotFound

[krallin]

This produces output like this:

INFO[2018-01-10T13:49:13+01:00] starting                                      iteration=0 job.command="sleep 2.5" job.position=0 job.schedule="* * * * * * *"
WARN[2018-01-10T13:49:14+01:00] not starting: job is still running since 2018-01-10 13:49:13 +0100 CET (1s elapsed)  iteration=0 job.command="sleep 2.5" job.position=0 job.schedule="* * * * * * *"
WARN[2018-01-10T13:49:15+01:00] not starting: job is still running since 2018-01-10 13:49:13 +0100 CET (2s elapsed)  iteration=0 job.command="sleep 2.5" job.position=0 job.schedule="* * * * * * *"
INFO[2018-01-10T13:49:15+01:00] job succeeded                                 iteration=0 job.command="sleep 2.5" job.position=0 job.schedule="* * * * * * *"
WARN[2018-01-10T13:49:15+01:00] job took too long to run: it should have started 1.503185062s ago  job.command="sleep 2.5" job.position=0 job.schedule="* * * * * * *"

See it action with: make build && ./supercronic integration/timeout.crontab.

[fancyremarker]

Very nice!