Fix pclomp::GICP for humble

[MaximSmolskiy]

Fix #81

Ported important changes from PointCloudLibrary/pcl#3854

After these changes:
ROS_DISTRO=humble

root@feedb965e72d:~/colcon_ws# build/ndt_omp/align src/ndt_omp/data/251370668.pcd src/ndt_omp/data/251371071.pcd 
--- pclomp::GICP ---
single : 124.054[msec]
fitness score: 0.220292
final transformation:
   0.999918   0.0127417 -0.00144976    0.490933
 -0.0127527    0.999888 -0.00781302    0.119676
 0.00135004  0.00783087    0.999968   -0.025651
          0           0           0           1

[amadeuszsz]

@MaximSmolskiy thank you for this PR!

In 84d5c81 I added #include <set> which make this PR compatible with ROS Jazzy.

Regarding #73:

• Was fixing Jazzy build.
• Was kind of "hotfix" - didn't overload checkGradient and potentially always runs optimization loop till max_iterations reached.
• Can be closed in favor of this PR.

@SakodaShintaro could you please take a look on this PR?

[SakodaShintaro]

(cc. @YamatoAndo )

Thank you for the pull request. Please let me confirm the background of this PR.

Currently, ndt_omp has been integrated into the autoware_core repository.

https://github.com/autowarefoundation/autoware_core/tree/3101380f1dcf2a5d09f868413506693b82855500/localization/autoware_ndt_scan_matcher/include/autoware/ndt_scan_matcher/ndt_omp

Also, this ndt_omp repository has been removed from autoware.repos.

autowarefoundation/autoware#5390

Is it necessary to make this change in this repository instead of autoware_core?

[amadeuszsz]

@SakodaShintaro Can we do both? 🙏🏻 This PR does not break Humble distro. We use this repo in https://github.com/tier4/CalibrationTools/

[SakodaShintaro]

These changes look good to me.

In the future, could you consider switching the dependency of CalibrationTools from ndt_omp to autoware_core?

[amadeuszsz]

@SakodaShintaro I can't answer for now. First we will check if we can make CalibrationTools autoware*-agnostic. If so, we might prefer to use standalone ndt_omp. Otherwise we will just use it from autoware_core.

Thank you for your review! 🙇🏻‍♂️