Service failure streak detection

A health checker appends one line per probe: "<timestamp> <service> <status>", where status is "ok" or "fail". An incident page should fire when any single service fails n times in a row, counting only that service's own probes, in log order; probes of other services in between do not reset the streak. Given the log lines and n (n >= 1), return true if some service reaches n consecutive "fail" probes, and false otherwise.