core(lantern): add configuration for precomputed network analysis

[patrickhulce]

Summary
As discussed in the variance meeting earlier, this adds the ability to read in precomputed network analysis values and write them to a file. Nothing happens automatically for now it's up to the caller to do all the booking, etc to limit the complexity and difficult questions until we can validate how effective this is.

Related Issues/PRs
#6775

[exterkamp]

Such a quick turn around!

I was thinking, should the data be additive? Like we record the data each time the --precomputed-lantern-data-path flag is present, so the format would be more like:

{  
   "additionalRttByOrigin":{  
      "domain1": [1, 2, 3],
      "domain2": [4, 5, 6]
   },
   "serverResponseTimeByOrigin":{  
      "domain1": [1, 2, 3],
      "domain2": [4, 5, 6]
   }
}

So that we can "build" more accurate averaged measurements? So when reading the file we look at all the records we've seen and extract some average or median or something?

[patrickhulce]

I was thinking, should the data be additive?

Definitely! Great idea!

IMO, this could probably wait until we prove out how much of an impact it has on the metrics, and even long-term fancy median/avg/merging logic could potentially be the caller's responsibility. I see potentially very different cases for CI and LR for example.

[brendankenny]

I still have strong reservations about this direction in terms of usability, desirability of effect, principle of least surprise, etc etc, but this implementation is a nice one :)