Service recovery time

A status probe writes one line per check as "<ts> <status>", where ts is integer Unix seconds (chronological) and status is "up" or "down". Postmortems need time-to-recover: find the FIRST "down" line, then the first "up" line after it, and return the difference in seconds between those two timestamps. If the service never went down, return 0. If it went down and no later "up" line exists, return -1.