Inspector multiprocessing is very slow

[nikpau]

MP in the Inspector is suboptimal

The way multiprocessing is implemented in pytsa.trajectories.inspect.Ispector.inspect() is suboptimal if the data to inspect grows large. This is because the Inspector first splits the data to inspect into as many small chunks as there will be processes. Afterward, the smaller splits are sent to each worker, rendering the bulk of work send and receive tasks.

Idea

We could try shared memory for parallelization, however we must find a suitable way to transform the TargetShip objects to native ctypes.