Add configurability for weights associated with resource scores

[eric-czech]

Most data sources provide multiple pieces of information for any one evidence record that need to be combined to produce a score for that evidence that can then be aggregated across all records for an association (distinct combination of target + disease). For example, GWAS Catalog scores for evidence records are comprised of the sample size for a study, the p value for the association, and a confidence score describing the strength of association between variant and gene (there is also an LD score that is handled by the code but not present in the data?). These values are also cast to some 0-1 score and multiplied together. The relative weights of these values should be configurable.

This level of configurability will likely not be very useful without companion distribution statistics on the individual constituents to see which ones are currently dominating the others, but that analysis can come later.