Understand why 1D analysis is slower than nD analysis

[schillic]

This is at least true for ISS: With lazy inputs: 18.53 seconds (1D) vs. 0.35 seconds (270D) Without lazy inputs: 13.51 seconds (1D) vs. 11.97 seconds (270D)

So the lazy inputs seem to have a major impact (and the reason why is not clear to me), but even without lazy inputs I would not expect better speed.

Note that the efficiency of the sparse analysis cannot be the reason because in nD we can only exploit sparsity for a system with no dynamics.

[schillic]

The above was the property checking mode. The reachability mode with the same options instead takes minutes to run, and both bigger partition and lazy inputs make it slower.