If we could do overlay the android test trace with this data, it would be really helpful to see temperature trends throughout the run. We couldn't get continuous data because that would require too much communication back and forth with the phone. 

This would be tangential to the dashboard metric though. Since anything we add to that trace would be lost when going to swarming since the android test runner is what produces the trace. Whats the timeline on moving android to swarming? I could get the temps added to the trace fairly easily I think, but if we are going to move to swarming tomorrow its not worth it.

We probably try go get Android swarmiing working in a few weeks, now that the memory issues are almost solved there. 

For now, I am inclined to just go with the easiest option, which is adding the temperature metric to the dashboard.

I think we can just add these 2 metrics (per benchmark):
experimental_device_tempreature_before_benchmark_run
experimental_device_tempreature_after_benchmark_run

Another option is we can the two metrics per story level:
experimental_device_tempreature_before_story_run
experimental_device_tempreature_after_story_run

I am leaning toward (2) as it's probably easier to implement & allows finer grain monitoring. Wdyt?

I vote for option 2 as well.

Stephen: the way to add this should be similar to how you add the benchmark_duration. Except this time, we add it per story, so you want to put before & after in https://github.com/catapult-project/catapult/blob/master/telemetry/telemetry/internal/story_runner.py#L237

Randy can probably help with making an API to query device temperature.

Not really working on this anymore.