#30 Add 'Coverage' to model statistics

[Peter Andrews]

This is the ratio of rows looked at to rows that we could make a classification for. Normally this is 100%. In the case where TIE is set to UNKNOWN, though, if there are ties then this ratio can be lower. This is important to know because you might have apparently great accuracy but in reality you made very few predictions. You could see this on the Best-Model tab because we give you the numbers for each attribute combination but this level of information is not provided for the the CV tests.