[core] Lazy import of pydantics

[rickyyx]

Why are these changes needed?

Closes #41338

The hypothesis is that we are importing some additional things when initializing the python workers, and changing the import to lazy seems to fix the issue.

See https://buildkite.com/ray-project/release/builds/2492#018c1923-c5d0-4e09-a67e-b45cf2c3b553

Master: tasks_per_seconds = 430
This PR: tasks_per_seconds = 530

Related issue number

Checks

• I've signed off every commit(by using the -s flag, i.e., git commit -s) in this PR.
• 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 added any new APIs to the API Reference. For example, if I added a
    method in Tune, I've added it in doc/source/tune/api/ under the
    corresponding .rst file.
• 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 :(

[rickyyx]

https://buildkite.com/ray-project/release/builds/2492#018c1923-c5d0-4e09-a67e-b45cf2c3b553 release test perf back to 530 level.

[rkooo567]

Q: consider adding a unit test to avoid regression? (i.e., import serialization_addon.py and make sure libs are not imported

[rickyyx]

Q: consider adding a unit test to avoid regression? (i.e., import serialization_addon.py and make sure libs are not imported

Sure - but i think what's more important is how we prevent this from happening again. Don't we have tests that track worker start-up time?

[rickyyx]

test it failed the test on master.