Delta tracking

[jpmorgan98]

Adding a bunch of new tracking features to explore various kinds of hybrid delta tracking schemes. Most of the meat and potatoes is in main.prepare where the majorant is computed and kernel.move_to_event where delta tracking actually happens.

Note that I am adding a bool to each particle to keep track of weather or not it is undergoing delta tracking. This feels like kinda a big deal

New Features

• A collision estimator
• Functions to compute both the macroscopic and microscopic majorants
• Ability to plot the majorant
• Rejection sampling
• Visualizer improvements with the ability to save time outputs as .gifs and print legends
• Hybrid methods including an energy cut off and material exculsion
• Adding documentation functionality to use

Some Todos

• Make delta tracking on/off logic in kernel.move_to_event a little more streamlined
• Make a separate compute_majorant functionality to plot the majorant without having to run the simulation
• Giving materials input-able names so they can be easier to plot and identify
• Allowing multiple material regions to be excluded (right now it's only 1)

Verification

Passing verification sueit, also producing the correct results with C5G7, C5CE, Kobyashi, and Dragon all of which see speedups when delta tracking is enabled. Or hybrid delta tracking.

New dependency

I am adding a new dependency so that the visualizer can make the movie. Its based on matplotlib and is pretty light weight.

Limitations

This delta tracking does not support fission tallies so also is not supported in k-eigenvalue computations

[ilhamv]

Thanks, @jpmorgan98. Note that in k-eigenvalue, we can use the collision estimator to get the estimate of the next iterate of k.