fix: Fixed bug where group folds and sample weights couldn't be used together

[dannycg1996]

Why are these changes needed?

Previously there was an error with using sample weights and group folds together (see related issue). This PR fixes that issue, and adds a test to ensure that this stays fixed in future. I also noticed that we were importing numpy twice within test_splits.py, so I removed the defunct import statement.

Related issue number

Closes #1396

Checks

• I've used pre-commit to lint the changes in this PR (note the same in integrated in our CI checks).
• I've included any doc changes needed for https://microsoft.github.io/FLAML/. See https://microsoft.github.io/FLAML/docs/Contribute#documentation to build and test documentation locally.
• I've added tests (if relevant) corresponding to the changes introduced in this PR.
• I've made sure all auto checks have passed.

[dannycg1996]

Hi @thinkall,
All of the tests seem to pass, except for those on the ubuntu 3.10 and ubuntu 3.11 builds, where test_gpu fails. Having looked at some of the other open PRs, such as #1402 and #1392, it seems that this test always fails.
Is this a test we can just ignore? If not, do you have any ideas on how to fix it?
Thanks for any help!

[thinkall]

Thank you @dannycg1996 for the PR! LGTM.