227 weights generalization improvements

[em-t]

A suggestion for how generalization could be handled for the cumulative inputs. Closes #227 but still requires some feedback.

If generalization is unsuccessful during weights calculation, return the weights table without the generalized columns (class, weight and std) and inform the user with a warning (instead of raising an exception and terminating the function).

Make a public function for performing classification and weights calculation for cumulative data, which can be called on a weights table to reclassify the data. Include different options for generalization strategies. (The different options that I provided for the generalization may not make sense. I'll remove any that are unnecessary and make any other modifications requested. Feedback from someone familiar with the method is welcome!)

Code-wise, this can probably be cleaned up some more. It's also noteworthy that this PR introduces custom warnings to the toolkit, which may not be something we want.