Retry failed mutations on new session at least once

[tjchambers]

On failed mutation runs the mutator reverts to a single job (-j1) to retry the mutation run in case the failure is a concurrency issue. If THAT run fails it moves on.

If you rerun all mutations, the failed mutation runs try again, but they do so IMMEDIATELY with -j1 instead of retrying with more cores.

IMO the first time in a run that filed mutations are re-tried they should retry without the job parameter. The current workaround is to just delete the log file, which will start the mutation attempt over.