Incorrect zeros included in emsd average when particles are dropped

[vivarose]

Have you ever noticed that trackpy imsd functions always look like they drop to zero at the longest lag time?

They shouldn't, but they do.

This zero is included in the emsd average. This appears to be incorrect.

I don't think this is particularly rare. The walkthrough shows many imsds that drop to zero. Each of them would pull down the emsd incorrectly.

I have posted a short jupyter notebook that shows the issue here: test_trackpy.ipynb

[vivarose]

As I wrote in my pull-request, now that I've fixed the issue, I find that it didn't make that big of a difference on my data. But I still can't stand inaccuracies, so I plan to use my fixed version going forward!