Suspected rounding errors in geodetic coordinates due to float32 digit limit

[bartboogmans]

Issue adressed by @eliatrevisan

It is suspected that the digit limit of numpy float32 is too little to capture reasonable motion in the timesteps of the simulation.

To do:

• Investigate the magnitude of rounding errors
• Adapt float types in respective places
• Check if errors went down.
• Evaluate performance impact

[bartboogmans]

The following branch is used to implement this. I accidentally did it in the branch that was also in progress to change the input arguments towards parameter format. I'll keep this as a combined workpackage for this branch.

https://github.com/bartboogmans/TurtleBoat/tree/ros2_param_implementation