Add visualization of execution of the computation graph over time

[kszlim]

Description

It would be useful to have an optional way to visualize the execution of the compute graph over time (and specfically see where your code is bottlenecking, how many cores is the work being scheduled on at a time, etc.

This could start pretty simple, but I imagine it could eventually be fairly sophisticated.

It would be helpful for both polars developers and users of polars (who want to optimize their code, for example, they might see they're doing too many with_columns, limiting the number of potential independent computations.