Reworked and Stable Sonar Object/Wall Detection

[AnthonZh]

Summary

• Added sonar_object_detection.py for a pipeline to process raw sonar scans, segment them, and retrieve normal angles and poses
  • Primarily use the Fourier Transform to denoise images
  • Segments the image and throws out segments that don't meet certain criteria
  • Uses orthogonal regression to retrieve the normal angle
• Updated and cleaned the sonar.py file to move utility methods to the appropriate file
  • Moved most calculation methods over to the utility file
• Add a raw image publisher to that both the raw sonar scan and the processed sonar scan are viewable in foxglove
• Update dockerfile to install scikit-image

Testing

• Tested with empty data and previously collected data
• Tested on land

[nathanaelren]

Converting to draft as has not been pool-tested yet.

[nathanaelren]

Started to go through it then realized y'all gotta lint :) Run python3 lint.py in the robosub-ros2 folder

[AnthonZh]

Should have fixed the issues, will mark as ready for review after a pool test.