bulkio: Incorrect transaction semantics in job_scheduler

[miretskiy]

Job scheduler does not use transactions correctly, and this can lead to deadlocks.
In particular, the query to find schedules to run does not use transaction; yet the updates
to schedules are done under txn. This can lead to a stuck process if the transaction retries
after any changes have been made to any of the schedules.

In addition, currently, a single schedule with any sort of error (planning, etc) would lead to the
entire scheduler loop to be undone.

[blathers-crl]

Hi @miretskiy, please add a C-ategory label to your issue. Check out the label system docs.

While you're here, please consider adding an A- label to help keep our repository tidy.

_{🦉 Hoot! I am a Blathers, a bot for CockroachDB. My owner is otan.}