Explore Ryan's idea about adaptively bumping predictions up/down based on coverage

[dshemetov]

Need more details here @brookslogan.

[brookslogan]

See this paper for a broader framework or Ryan's research presentations about the simpler application of bumping up or down in event of under/overprediction. See also the related note under this issue to bring this feature to the tooling side.