Improve performances when events are used

[luyckxn]

Several improvements may be done:

-> Don't reexport the model to mrgsolve/RxODE at each iteration -> Cache the mrgsolve/RxODE model if it does not change from iteration to iteration -> Investigate method 'mrgsim_q' in mrgsolve for repeated simulations

[AndreasCalvagone]

Or maybe something similar to the dynamic dosing feature in mrgsolve: https://mrgsolve.org/blog/posts/2024-new-1-4-0-evtools.html