Enable the benchmark function to have state as input and output

[Simpag]

Allowing the benchmark function to accept states as input and returning states as outputs would allow users to save the current state and continue progress at a later point in time. This also enables users to create their own evaluations which are not necessarily benchmarks such as plotting image samples.

The state in this case would be a distributed algorithm with a corresponding network. I would suggest using the same structure as the output from benchmark._run_trials which is dict[DstAlgorithm, list[Network]]. A helper function could be made to initialize new runs.