Gather info on Desktop probes that are being sampled and have low client counts

[edugfilho]

We sample Windows + Release Desktop probes at 10%.
As part of an effort to understand how reliable the data we show on GLAM is, we want to know how many of the sampled probes have low client counts, because sampling from a low client count makes the aggregated data lose its representativeness.

What is low? baby don't hurt me
Probably something under hundreds of thousands. But perhaps a more informational outcome of this will be a comparison between the lowest counts vs the average counts.