Performance advice for Jump simulation tutorial

[TorkelE]

Adds a single tutorial, providing some advice on Jump simulation performance:

• Managing of solution saving using save_positions and saveat.
• Types of jumps and why e.g. k, E + S --> E + P is better than k*E, S --> P.
• How to select a good solver and why e.g. RSSA is probably better than Direct (and RSSACR should be preferred for big models).
• Mention of the existence and principle of hybrid simulation, however, notes that these are currently only a work in progress.

[TorkelE]

Also, @isaacsas, If you know a good reference book/article on Jump simulation it would be useful to know (so I could put a reference to it here, for further reading).

[isaacsas]

I think for bio readers this is a nice one that is focused on algorithms: Luca Marchetti, Corrado Priami, Vo Hong Thanh,
"Simulation Algorithms for Computational Systems Biology".

[TorkelE]

I tried writing the tutorial assuming that someone wasn't familiar with rates, propensities, and rate laws, and hence basically just used rate instead of propensities. I don't min making it more rigorous, but then I want to add a proper introduction of what these things are. I think if we just start using propensity people not familiar with it will get confused really quickly.

(long term I want a better introduction to this somewhere in the docs as well, just so that it can be cross-referenced in situations like this)

[ChrisRackauckas]

Is it worth mentioning small case static array stuff?

[isaacsas]

We should probably benchmark that case and see if there is a speed up. I don’t think we’ve ever done so.

[ChrisRackauckas]

I have, you can see it here:

https://github.com/sdwfrost/Gillespie.jl?tab=readme-ov-file#benchmarks

It's a bit old but it works well for small problems. In fact, for the smallest of small problems we should probably make a different codegen approach, but that's a whole different story.