Add DNS measurement analysis code

[ben-jones]

Add code to process DNS primitives and determine if censorship is happening

For each measurement, we will do the following:

• check the resolved IPs against control measurement (add whether the control was consistent)
• check the ASN of resolved IPs against control measurements (also add whether the control was consistent)
• check the server header of resolved IPs (conduct an HTTP GET to the IP without a Host field or path)

For each country, we will:

• check if different domains resolve to the same IP (also indicate if the control measurements do the same)

@rpanah, @feamster, @phillipa, @jakubd, please let me know if you can think of additional analysis steps.

[phillipa]

The appendix of my TWeb paper has a heuristic I used to identify DNS redirections using the control data.

It's a bit confusing to read but the idea is that if you have N sites that resolve to IPs in M ASes in the control data, and these N sites resolve to a single IP in the field, then it is likely that this IP is a DNS redirection.

Also easy stuff like looking for bogons/non routable IPs returned.

Some of this analysis can be done offline in the data management code too.