new functionality: allow facilities with material losses (path_split) to either lose a portion of whole components or a fraction of each component's mass

[rjhanes]

Lower priority - not strictly needed for the glass study

Follow on from #208

From that issue:

An alternate implementation here would be to use the loss fraction as the probability on a Boolean random variable, and making it so instead of 10% of each component going to landfill, 10% of whole components go to landfill and the rest continue as is. This might be an adjustment to make in the future and would help to capture e.g. window breakage during installation with a little more specificity.

[rjhanes]

Note: Scenarios in which the component_scaledown parameter are used should NOT make use of this functionality (when implemented). When modeling 100s or 1000s of components, it does make more sense to apply the path_split logic as originally implemented rather than a Boolean.