Batch runner params

[dcunning11235]

PR that I think satisfies #607

Added two new utility classes (ParameterProduct, ParameterSampler) for creating parameter combinations, and a base class FixedBatchRunner, which is now the super for BatchRunner, that takes a list(-like) of parameter combinations to run.

[coveralls]

Coverage increased (+0.4%) to 76.814% when pulling fa7bbb0 on dcunning11235:BatchRunner_params into 2359a12 on projectmesa:master.

[coveralls]

Coverage increased (+0.4%) to 76.814% when pulling fa7bbb0 on dcunning11235:BatchRunner_params into 2359a12 on projectmesa:master.

[jackiekazil]

@dcunning11235 ty for the PR. We have a few complex issues happening at the moment. We didn't forget about you. It may take us a few days to get to review/discuss this. Hang tight!

[djfurman]

looks good! @jackiekazil - review complete; nothing blocking added #676 to backlog for only enhancement I saw.

[jackiekazil]

Great!

[nathanvdlei]

Hi all! I have been trying to use this logic with multiple variable_parameter entries. For this we use two lists, where one starts at 0.5 and the other starts at 1:
variable_params = {'clustering_std': list(np.arange(0.5, 4, 0.5)), 'avoid_fights_factor': range(1, 11)}
However if we call get_model_vars_dataframe() we seem to find that the wrong list was assigned to the wrong variable. [NOTE how 'clustering_std' is supposed to start at 0.5 and 'avoid_flights_factor' at 1. In the picture we see this inversed.

Investigating ParameterProduct and ParameterSampler we have not been able to identify the issue yet. The picture attached demonstrates the possible bug. It might have to do something with sorting in some iterable or list.