Bug 272578

Version:           2.0.3
OS:                Linux

There are a few issues with point density:
- the low setting sometimes results in far too few points being drawn
- the medium setting has strange intervals
- maybe other issues...

Reproducible: Always

Steps to Reproduce:
Load the attached test file, to be used with the TWIN1 sample data file in the source tree. 
1) The medium setting has strange intervals (particularly on the left part of the plot)
2) Change the green curve's density to "Low": you get only one point!
3) Zoom maximum with medium: you get almost no point!


Expected Results:  
- The points should be evenly spaced (with the question of what to do when the x data points aren't evenly spaced: try to have the points evenly spaced in terms of abscissa, or in terms of number of samples? I prefer the first option) 
- There should be a few points with the low density in all cases, unless we've zoomed in so much that there really is only one data point left in the viewport. I think density should relate more to the number of visible points on screen than the number of samples skipped between two points. The rule could be something like:
- all: clear
- high: strive for 20 points for the curve (number possibly dependent on plot width in pixels?)
- medium: try to get 10 points
- low: 5 points

I think it'll take some experimenting to get it right, possibly we could remove one level. But it can't stay as is.

OS: all