Do weighting for prediction refinement in resolution shells
Migrated from Jira (CRYS-152). Prediction refinement can probably be improved as follows:
Currently: It puts the strongest reflections nearest to Bragg, i.e the low-res reflections go nearer Bragg than high resolution ones. Better: It should put the strongest reflections at a particular resolution nearest to Bragg, thereby weighting a bit more evenly across resolution shells. May need some thought about how the weighting should appear across resolutions.