[Fix] Fix unit test of datasets

[JingweiZhang12]

Thanks for your contribution and we appreciate it a lot. The following instructions would make your pull request more healthy and more easily get feedback. If you do not understand some items, don't worry, just make the pull request and seek help from maintainers.

Motivation

Modification

1. Fix unit test of datasets by regenerating new annotation files.
2. Reduce cuda memory in UT.

BC-breaking (Optional)

Does the modification introduce changes that break the back-compatibility of the downstream repos?
If so, please describe how it breaks the compatibility and how the downstream projects should modify their code to keep compatibility with this PR.

Use cases (Optional)

If this PR introduces a new feature, it is better to list some use cases here, and update the documentation.

Checklist

1. Pre-commit or other linting tools are used to fix the potential lint issues.
2. The modification is covered by complete unit tests. If not, please add more unit test to ensure the correctness.
3. If the modification has potential influence on downstream projects, this PR should be tested with downstream projects.
4. The documentation has been modified accordingly, like docstring or example tutorials.

[JingweiZhang12]

I guess sample_id in two places: https://github.com/open-mmlab/mmdetection3d/blob/dev-1.x/mmdet3d/evaluation/metrics/kitti_metric.py#L419 and https://github.com/open-mmlab/mmdetection3d/blob/dev-1.x/mmdet3d/evaluation/metrics/kitti_metric.py#L511 don't need to be replaced by sample_idx ?

[JingweiZhang12]

The package of waymo-open-dataset-tf-2-1-0 does not support the env of python3.9 and python3.8.

The waymo-open-dataset-tf-2-4-0 has some dependency conflicts with lyft-dataset-sdk, such as typing_extensions, importlib-metadata. It can't be solved in CI.
Suggest not to install waymo-open-dataset-tf in the env of CI.