Add benchmark for time of computation scaling wrt chain len

[loscati]

It would be nice to have a scaling test to verify how the performances scales with chain len.
An average over multiple and different chains, at least 3, would be nice.

Since the probability of entanglement formation increases with length, one might consider a normalization to keep track of this effect.