feat(autoware_image_distortion): correct camera image distortion

[sakunaka39]

Description

This PR introduces a new package autoware_image_distortion to rectify distorted images from cameras. This package subscribes to raw or compressed images and camera info, performs rectification using image_geometry, and publishes the rectified images.

Changes

• Added autoware_image_distortion package.
• Implemented ImageDistortion node to rectify images using image_geometry::PinholeCameraModel.
• Added launch files for single camera and multi-camera configurations (e.g., aip_x2_gen2).
• Supports both raw and compressed image transport.
• Added license headers and resolved pre-commit errors.

Results

Before (Distorted)

After (Rectified)

How was this PR tested?

1. Launch the node with a camera driver.
   ros2 launch autoware_image_distortion distortion.launch.xml sensor_model:=sample_sensor_kit
2. Verify that rectified images are published to /sensing/camera/camera*/image_rect.
3. Check the images using rviz2.

[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.

[manato]

@sakunaka39
Thank you for your contribution!
I left some comments that I noticed. I'd appreciate if if you could consider them.
Besides, from the perspective of the node behavior, I think autoware_image_"un"distortion is a more appropriate name for the node because this node performs undistortion (getting rid of lens distortion). Considering this point would also be appreciated!

[manato]

Would it be possible to add description for the interpolation_method argument, like other launch files?
Judging from the PR description, I guess this file will be the entry point, which users mainly execute. If that is the case, the description would be helpful.

[manato]

Suggested change

	pub_rect_.publish(out_img_msg, info_msg);
	auto out_info_msg = sensor_msgs::build<sensor_msgs::msg::CameraInfo>()
	.header(info_msg->header)
	.height(model_.rectifiedRoi()height)
	.width(model_.rectifiedRoi()width)
	.distortion_model(model_.cameraInfo().distortion_model)
	.d(sensor_msg::msg::CameraInfo::_d_type(model_.cameraInfo().d.size(), 0.)) // zero distortion
	.k(model_.intrinsicMatrix())
	.r(model_.rotationMatrix())
	.p(model_.projectionMatrix())
	.binning_x(model_.binningX())
	.binning_y(model_.binningY())
	.roi(model_.rectifiedRoi());
	pub_rect_.publish(out_img_msg, out_info_msg);

I think publishing the input camera info here is not appropriate, otherwise there will be two publishers that publish the same topic.
Besides, strictly speaking, camera info for the undistorted image would be slightly different from the original one.
My suggestion reflects this point. Since I haven't tested if it can be built, I would appreciate it if you test it in case you adopt it!