Schedule next job at a shorter interval, if the previous job failed

[jpmckinney]

Before this week, a failed job meant no other job would ever be scheduled (until the failed job was deleted). Now, a new job will be scheduled at the regular interval. However, some publications have longer retrieval frequencies. If a job fails, it can be a long time before the next attempt.

We can add logic to e.g. schedule a job if it's been a fixed amount of time since the failed job, e.g. switch to the min frequency of 1 month, instead of using the publication's frequency (unless that frequency is "NEVER").

Add relevant tests first #128