[optimization] add a way to log each function call and duration

[lesquoyb]

Describe the bug
While writing documentation about how to optimize models I noticed that in some case you have no choice but to log every action/reflex called and their duration to get an idea of where to look for optimizations.
This process is quite tedious and always the same so it's perfect for automating and I was thinking that we could provide to the users an execution mode that would automatically do it and export the results in a csv or something to be visualized in an external program

[AlexisDrogoul]

Isnt'it the purpose of the existing benchmark mode ?

[lesquoyb]

Benchmark only works for the code contained inside of it, I was thinking about an execution mode that would basically wrap every action/reflex by an equivalent of benchmark.
Another problem with benchmark is that it doesn't produce raw data and do not give access to it in other way than a text in the console (as mentioned in #329)

[lesquoyb]

So to give some updates on that, there's a benchmark mode on the experiment level, which is different from the statement I was thinking about.
This mode records all lines of code executed with the number of calls and accumulated duration.
I've added this information to the page I'm currently building on optimization so closing this issue, but I'm quite that if people actually use it, it will need to be refined a bit