Framework for follow-up experiments

[bassosimone]

Should we retry measurements directly after they fail? Does this mean a single experiment may return more than a single measurement? What other designs are possible?

[bassosimone]

I think I have sketched this out in github.com/ooni/netx. While we run measurements we automatically run quick follow-up experiments. The data structure we return from netx to probe-engine is a super-set of the OONI data model, where we include a "scoreboard" structure that contains information on all the oddities that we noticed, which of them we could measure on the fly, and which of them instead should be investigated later/by the pipeline.

Also, don't know why I didn't properly tag it in the first place, but this was clearly part of #87