Why are these changes needed?

This is a first step to handling ensuring that only output from the relevant job is sent to each driver. This PR will make each worker write its own output to a file titled worker-{worker-id}.out (which it already does), but use produced output from jobs will now be written to a file titled worker-{worker-id}-{job-id}.out. stderr will also be redirected accordingly.

Related issue number

First of 2 steps towards closing #4029.

Checks

• I've run scripts/format.sh to lint the changes in this PR.
• I've included any doc changes needed for https://docs.ray.io/en/latest/.
• I've made sure the tests are passing. Note that there might be a few flaky tests, see the recent failure rates at https://ray-travis-tracker.herokuapp.com/.
• Testing Strategy
  • Unit tests
  • Release tests
  • This PR is not tested (please justify below)

No tests here. In some ways this is just a non-trivial refactoring in the sense that it shouldn't effect end behavior, thus it's sufficient to not break anything. When behavior changes (i.e. next PR where output is actually sent to specific drivers only), test cases will be included.