[CI] Deflake test metrics

[simon-mo]

Closes #19715

Verified no longer flaky by running it in build box environment. The dashboard takes more than 2s to startup. But the test wait_until_succeeded_without_exception timeout after 1s.

Why are these changes needed?

Related issue number

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/master/.
• I've made sure the tests are passing. Note that there might be a few flaky tests, see the recent failures at https://flakey-tests.ray.io/
• Testing Strategy
  • Unit tests
  • Release tests
  • This PR is not tested :(

[jiaodong]

thanks for the quick fix ! Do we know what contributed to the potential increase of dashboard starting time ? I think my PR is simply deleting existing unused job server and in its async run() it will create an JobManager instance.

Maybe calling ray.remote() on constructor ?

https://sourcegraph.com/github.com/ray-project/ray/-/blob/python/ray/_private/job_manager/job_manager.py?L166:14

[simon-mo]

Probably yeah.