feat(auto_labeling_3d): add tool to convert info to annotation tool format

[Shin-kyoto]

Summary

Add tool to convert info to annotation tool format

• I added the code to convert info.pkl to deepen.json format.
• We have the plan to add conversion tool from info.pkl to segments.ai format.
• With this PR, we can output the annotation tool format pseudo-label. By using this, we can use deepen_to_t4 in tier4_perception_dataset

Change point

• feat(auto_labeling_3d): add tool to convert info to annotation tool format

Note

This does not influence other projects.

Test performed

I confirmed that this PR works correctly by checking this PR with running the following command.

root:/workspace/tools/auto_labeling_3d/create_annotation_tool_format# python create_annotation_tool_format.py --input /workspace/work_dirs/auto_labeling_3d/pipeline_example/pseudo_infos_centerpoint+bevfusion+streampetr_filtered_tracked.pkl --output-dir /workspace/work_dirs/auto_labeling_3d/pipeline_example/ --output-format deepen
Load pseudo label from /workspace/work_dirs/auto_labeling_3d/pipeline_example/pseudo_infos_centerpoint+bevfusion+streampetr_filtered_tracked.pkl
Generate deepen format json: /workspace/work_dirs/auto_labeling_3d/pipeline_example/Pseudo_x2_dev_b3902d62-9777-496d-846b-39c3db7b9dcf_2025-04-09_11-34-42_11-35-02.json