r isn't uniformly distributed over the annulus

[TuurDutoit]

I got this email from someone:

I just felt I should mention that your implementation of bridson has a bug.
You select the candidate from the annulus using a simple polar coordinate.
Sadly, 'r' isn't uniformly distributed over the annulus. Fortunately, r^2 is.
Selecting with
r = sqrt(random(r*r, r2*r2));
will fix the bug.
If you plot an annulus with your existing samples, it'll have a lot of points near 'r'.
if you plot with the fix, the annulus is properly uniformly distributed.

Figure out what he means and fix the problem.

[fenwick67]

When you have a circle or ring that you sample in polar coordinates (that is, you pick a random angle and random radius), the points are more bunched together toward the center.

codepen example:

http://codepen.io/fenwick/pen/qqQzLm?editors=1010

This is the line you need to change:

https://github.com/TuurDutoit/bridson/blob/master/index.js#L105

[TuurDutoit]

Yeah, that's what I thought, just haven't found the time to fix it yet. I'll do it as soon as possible.
Thanks for the example, though!

[TuurDutoit]

Implemented a fix in a new branch better-distribution. Can you check if that's correct?