Replace BNO node with new IMU node that fuses 085 with LSM

[yameatmeyourdead]

In implementing the cascaded pid controller for the updated flight controller we require fused IMU data from our two sensors.
Bug Eric about this, probably need a Kalman filter of some kind or use a pre-existing implementation.
Would also like to publish the raw data from each sensor too.

ROS has standard message type for IMU data that includes covariance matrices:
sensor_msgs/msg/imu

[yameatmeyourdead]

Consider using
https://github.com/xioTechnologies/Fusion
and a weighted sum of overlapping sensors?

Has the potential to cause problems under extended duration accelerations as it is a complimentary filter and not an EKF. But the system dynamics here are really narsty for implementing a kalman filter...

[Hermanoid]

https://docs.ros.org/en/melodic/api/robot_localization/html/state_estimation_nodes.html
There's a good chance that the "default" ROS2 EKF filter's "omnidirectional motion model" would probably do okay-ish for our underwater situation. "Just add some VSLAM and you're chilling" or something.
I don't see either of these filters doing great, though, with only these two sensors. This means we should probably choose whichever node allows us to reset our velocity to zero manually when starting a transverse, because even 30 seconds of error integration might be beyond our ability to accurately measure.