Next version of libra live

It's time for a new version of Libra Live!

In preparation for [SatSummit](https://2018.satsummit.io/), we're building a new version of Libra Live to demo at the event.

The goal of the demo (poorly paraphrasing @ianschuler):

- In previous demos the story was "here's what's possible if data is stored in an accessible way."
- This time the story we're telling is "USGS is now sharing data in an acessible way, and look how cool it is."

Our technical goal is to create a demo app using react native and aws lex that allows users to give voice commands that query a real-time satellite image tiler that provides Landsat ARD tiles and display the tiles on a map.

## Scope of work

### Tiler & data
- [x] get buy-in from the landsat folks to use the california data for this presentation @scisco / @ianschuler
- [ ] ask digital globe if they are willing to make data available for this presentation @ianschuler
- [ ] move landsat data to a public bucket @scisco
- [ ] deploy [marblecutter-ard](https://github.com/developmentseed/ard-backend/) on the devseed aws account @scisco

### Mobile App
- [ ] reimplement [libra live](https://github.com/developmentseed/libra-alexa) as a react native app
- [ ] create simple design (probably based on existing version of libra live, but for phones)
- [x] use aws lex api instead of the alexa api for voice chat interaction
- the existing [slot types, intent schema, and utterances](https://github.com/developmentseed/libra-alexa/tree/master/skill/config)
- [x] capture audio to send to aws lex
- some existing react native plugin options:
- https://github.com/jsierles/react-native-audio
- https://github.com/kevinresol/react-native-sound-recorder
- [x] use [react-native-mapbox-gl](https://github.com/mapbox/react-native-mapbox-gl) as the map that renders tiles from marblecutter-ard
- [ ] make both android and ios versions available
- [ ] allow users to install the app using [TestFlight](https://developer.apple.com/testflight/)
- [ ] add a `show high-resolution images of {location}` command using Planet Labs california data

### Machine Learning
- [ ] create plan for involving machine learning in the demo @drewbo
- it should be simple, like "highlight avocado crops"
- [ ] implement an api endpoint either on marblecutter-ard or as a standalone service

## Timeline

SatSummit starts on September 19. We should have a prototype at least a week before that, by September 12.

### Rough milestones

#### August 31
- verify that capturing audio on react-native and communicating with aws lex works well
- get in touch with USGS and DG about using data
- implement initial map that communicates with the marblecutter-ard tiler

#### September 7
- implement all the lex commands the app needs
- make design improvements as needed
- implement machine learning endpoint and voice command

#### September 12
- working demo ready to test and debug
- deploy using TestFlight