adjust DELTA_T_SI depending on SIMDIM to improve the CFL condition

[PrometheusPi]

This pull request adjusts DELTA_T_SI to improve the Courant-Friedrichs-Lewy condition depending on the dimension of the simulation SIMDIM. It closes issue #763.

For 2D we now get:

PIConGPUVerbose PHYSICS(1) | Courant c*dt <= 1.00204 ? 1

For 3D we now get:

PIConGPUVerbose PHYSICS(1) | Courant c*dt <= 1.00556 ? 1

[ax3l]

As I proposed earlier, I would like to use the same time step for both 2D and 3D. That simplifies a lot of work, e.g., both simulations have the same time scale regarding to steps...

The "harm"/not-super-ideal CFL in 2D is ok for that case, it's an example that should be easy to read and to compare with, it is not a PhD thesis ;)

[PrometheusPi]

I adjusted the set up (I missread your comment.):

now we get for 2D:

PIConGPUVerbose PHYSICS(1) | Courant c*dt <= 1.23155 ? 1

[ax3l]

great! 👍 thank you for the fast turn-around!