fix(lidar_centerpoint): non-maximum suppression target decision logic

[technolojin]

Description

1. fix non-maximum suppression target decision logic
   Before: If both of the object class are in the target label, caluclate IoU. If not, check distance and calculate IoU if the distance is less than a threshold.
   After: If the label pair is not the same and one of them is pedestrian, do not suppress. Then, check distance and calcuclate IoU only if the distance is within a threshold.

2. optimize non-maximum suppression search range
   Keep the squared value of the distance threshold.

3. remove NMS type
   Since there is no other NMS option, remove the logic and parameter to select NMS type.

Related links

NMS PR: #1935

Issue:

• pedestrian near a car is not detected Misdetects the pedestrian near a car #7563

How was this PR tested?

Performance test https://evaluation.tier4.jp/evaluation/reports/16239dd4-5213-5843-ada0-49341293bb6d?project_id=prd_jt
Execution test https://evaluation.tier4.jp/evaluation/reports/26db2be4-4fe5-5fe9-8221-320a88f318d7?project_id=prd_jt

Notes for reviewers

Remove unused parameter autowarefoundation/autoware_launch#1267 will be merged after this PR.

Interface changes

Parameter iou_nms_target_class_names will be removed.

Effects on system behavior

None.

[github-actions]

Thank you for contributing to the Autoware project!

🚧 If your pull request is in progress, switch it to draft mode.

Please ensure:

• You've checked our contribution guidelines.
• Your PR follows our pull request guidelines.
• All required CI checks pass before marking the PR ready for review.

[kminoda]

@technolojin Would you separate the PR for transfusion? I want to make the PR as small as possible (OK for image_projection_based_fusion and centerpoint as those two packages are dependent)

[codecov]

Codecov Report

Attention: Patch coverage is 87.50000% with 1 line in your changes missing coverage. Please review.

Project coverage is 29.59%. Comparing base (9553414) to head (8d9311a).
Report is 11 commits behind head on main.

Files with missing lines	Patch %	Lines
...rpoint/lib/postprocess/non_maximum_suppression.cpp	87.50%	0 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #9595      +/-   ##
==========================================
- Coverage   29.59%   29.59%   -0.01%     
==========================================
  Files        1442     1443       +1     
  Lines      108512   108514       +2     
  Branches    41409    41405       -4     
==========================================
- Hits        32119    32115       -4     
- Misses      73269    73278       +9     
+ Partials     3124     3121       -3     

Flag	Coverage Δ		*Carryforward flag
differential	6.26% <87.50%> (?)
total	29.59% <ø> (-0.01%)	⬇️	Carriedforward from 9553414

*This pull request uses carry forward flags. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[technolojin]

@technolojin Would you separate the PR for transfusion? I want to make the PR as small as possible (OK for image_projection_based_fusion and centerpoint as those two packages are dependent)

@kminoda
Then, the maintainer of the transfusion will decide if the NMS logic to be fixed or not.

[kminoda]

@technolojin I'm OK with it.
(We had some PRs that changes transfusion and centerpoint at the same time, and this made the cherry-picking in some of our projects very difficult. My point here is to avoid such potential conflicts)

[kminoda]

LGTM