Use random generators and the possibility of reproducing results

[Boorhin]

Hi,
I have worked a bit on the random walk behaviour and the engines to generate random numbers in opendrift doesn't satisfy numpy recommendations anymore.
I am suggesting to create a self.rng=np.random.default_rng(seed=12345) The Seed sequence number could also be modified by a config parameter.

This is as simple as
elements.Dx= rng.normal(0,1,nb_timesteps)
elements.Dy= rng.normal(0,1,nb_timesteps)
elements.Dz= rng.uniform(-1,1,nb_timesteps_vert_mixing)