Dashboard and label assist changes

This is a companion to this phone release:
https://github.com/e-mission/e-mission-phone/releases/tag/v3.2.3

Let's create a release before we start making the DocumentDB-oriented changes

What's Changed

• Change the geojson export code to use confirmed trips instead of clea… by @shankari in #837
• Check in some additional utility scripts by @shankari in #836
• Support additional types of push notifications by @shankari in #840
• Add support for generating metrics based on confirmed trips by @shankari in #841
• Fix label assist scripts by @shankari in #842
• Fix median_speed call when returning confirmed trips by @shankari in #843
• Make sure that the mode label fields are prefixed by label_ by @shankari in #844
• Replace median_speed by mean_speed by @shankari in #845
• Create a new function to expand final labels by @shankari in #846
• Create a dataframe version of has_final_labels by @shankari in #847
• Create a script that exports user travel diaries as geojson by @shankari in #848

Full Changelog: v3.0.0...v3.2.3

Add basic support for auto-labeling trips

• Server side changes for infinite scroll (#795)
• Multiple auth policies for aggregate calls (#801, #802)
• Build models of common trips and use them for auto-labeling (#829)
• Include the auto labels into the pipeline as the inferred_trip field of confirmed trips

Add support for confirmed trips

• Add a configurable result limit
• Import the database module correctly
• removed unused environment files
• support language-specific surveys
• fix tour model generation code
• fix some dependencies

Point release with accumulated fixes over the year

This is also likely to be the the last release from this branch before we switch over to the GIS branch as the default.

The biggest changes were:

• migration to the most recent version of anaconda
• migration to the most recent version of all the support libraries, and
• upgrades to the setup scripts,
• listing all dependencies,
• removing all deprecated code, and
• adding an automated test pipeline

Other changes include:

• switch to the most recent version of anaconda
• added more simulation code
• add a new method to store data synchronously
• use it for the destination_confirm class
• minor fixes to openid auth server validation
• add support for routing with OSRM
• fixes to push notifications + supporting FCM on iOS as well
• add a script to upload a timeline to a remote server
• fix performance issues with reading incident entries from the usercache
• upgrade the motionactivity formatter since we no longer use the internal android class names for the fields

Performance improvement + support for new google play services encoding

• Performance improvement (99%!) #638
• Update motion activity inputs to support the new google play services version (11.0.1) #642
• Minor fix to remove obsolete script

(again) Bump up version number to address docker improvements

More changes to the environment -> new docker image -> bump up the version number for consistency.
This is primarily due to the changes in #637

Bump up version number to address docker improvements

Thanks to @alvinalexander we can now use docker-compose with environment variables to configure the docker image. But that requires minor changes to the startup script (#618), so we need to publish a new image.

Let's bump up the version number very slightly for proper versioning.

Support the creation of docker images + fixes to load/store scripts

• support exporting non-anonymized data for a particular study by directly specifying the study name
• support loading unprocessed data that was downloaded from the usercache, so it doesn't have the formatted time expanded out. This data needs to be reloaded into the usercache and not the timeseries

• minor readme fixes
• minor client setup fixes to customize the base app

Add support for creating and using the decision tree for mode inference

This change was originally started back in Jun 2017 by @jensteff But finally, the pipeline is ported over and we can do mode inference again! This uses the data collected from the Transportation Sustainability class as a base, and uses a decision tree for mode inference.

Other minor changes include changing the list of requirements to support windows (from @omrachel)

python3 + multiple fixes

• Refactor auth code into its own module
• Switch from ionic push to firebase push
• Move out a bunch of flaky/currently invalid tests
• Bunch of code cleanup
  • moves-style data
  • clients code
  • analysis
  • move exploratory notebooks to their own repository
• python2 -> python3
• handle conversion to pymongo3
• make it easier to set up a well-defined server environment
• support multi-tier setup, including db auth
• fix tests to handle minor python3 changes
• switch to new jupyter notebook + new folium
• zephyr documentation
• prepare to create an open data server
• move all client stats to the timeseries
• post-deployment fixes
  • pipeline reset fixes
  • fix failing test cases from open data server
• match ground truth to trip
• final fixes for the ionic -> firebase conversion