reproducable results indepent of `n_jobs`

[selmanozleyen]

Description of feature

This feature would mean we wouldn't create same results as old versions but it would fix:

• ripley reuses same seed for every simulation once a seed is provided #1232
• Would not use sequential seeds (which might cause correlated random results)
• Threading changing the results reproducibility

Would be also the best way to switch to numba since we could expect same results when numbafied also (since each seed is determined by the source seed)