Clump users with relatively low counts into a single "(other)" bucket.
This potentially cuts down the number of data sets for the bar graph
way down, and therefore increases the odds that it'll be possible to
show the per-user bar graph.
Currently the criteria for clumping a user are:
1) No changes/edits in the last or current month (so pie charts will
never show "other").
2) No bar in the chart would be more than 1% of the overall chart
height.
1% is pretty low, so another pass at this might be to raise the bar
iteratively until only a certain number of users (TBD) are left
standing.