Modularize ROS2 packages and Docker images

[luca-byte]

We want to separate the code in different ROS2 packages (reseq-core, reseq-digital-twin, reseq-sensors) and split them on two different docker images: the first will use Ubuntu Jammy with ROS, the second for the sensors only will need to deal with the RealSense peculiarities. We alse want to be alble to start both the containers with the same ROS2 configuration which will be passed to both, and provide a default configuration.

• Most modules can use only the standard image (Ubuntu Jammy, arm64) like this (link)
• The optional RealSense module will need some extra work
  • Either compile from source (as it is now)
  • Or use NVIDIA Isaac ROS (preferably)
    • Specifically the RealSense page
    • Or the other pages that explain how the container infrastructure work link.
    • We are specifically interested in the image compression at the moment since we have a CPU/bandwidth bottleneck
• When splitting we are interested in separating also the ROS2 packages (link to EC’s task)
  • creating a dependency chain
  • and correctly identifying the dependencies for the single package
  • (core reseq, sensors, digital twin)