feat: dockerize the site

[mcdurdin]

Note: this removes the non-docker configuration, to reduce maintenance.

Development version runs in two containers in order to simplify live-reload. Production version will run in a single container.

See README.md for details of usage.

• TODO: update to published version of shared-sites when feat: support for status.keyman.com shared-sites#89 merges
• TODO: establish k8s pod before merging this

Test-bot: skip

[keymanapp-test-bot]

User Test Results

Test specification and instructions

User tests are not required

[ermshiperete]

Not directly related to this PR but triggered by it: I only skimmed over the changes and realized that I'm not clear how it's split up in two containers. Looking at the existing documentation I get an idea how it's probably done. I think at some point we should improve the readme and add a few sentences explaining what we understand here as server and client (server=backend running on the server, client=frontend running in the browser?) and which directories implement that (I guess since there is a server directory that probably implements the server part, and so probably public will have the client part, but I'm not sure...)

[darcywong00]

Why wouldn't this be in clean_docker_containers?

[darcywong00]

If we want to use status.keyman.com.localhost for local dev,
maybe we need to add an entry to

https://github.com/keymanapp/website-local-proxy/blob/master/resources/nginx.conf

[mcdurdin]

Note: I have changed base to staging (https://com-keyman-staging-status.azurewebsites.net/) while we test this, to make sure that the deployment doesn't break the main status.keyman.com site. I will follow up with a production PR once we are confident that the azure deployment still works.